{ "generated_at_utc": "2026-06-14", "count": 294, "cases": [ { "id": "LIMEN-AUTH-000014", "title": "FTC Air AI complaint-TRO-stipulated-order package on conversational-AI business-opportunity marketing, now disposition-backed but kept non-core", "summary": "Official FTC complaint, TRO-motion, and March 2026 stipulated-order package on Air AI's conversational-AI business-opportunity marketing, now direct-document-backed through disposition on this host and usable as a bounded settlement-sidecar rather than a title-only or complaint-only queue row.", "authority": "authoritative_candidate_settled_noncore", "anchor": "SRC-AUTH-0027", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Finance & insurance" ], "system": "conversational ai;sales automation", "source": "", "caveat": "Use complaint-plus-stipulated-order posture together and preserve the order's non-admission clause; do not turn this into a hidden-human finding or a general AI-washing prevalence claim.; This row is now strong enough for a bounded non-core settlement sidecar on conversational-AI business-opportunity / human-replacement marketing governance, but it still does not outrank the authoritative core's sf08 court/public-record blocker.; The order text supports a business-opportunity ban and monetary-judgment posture; it does not itself prove technical capability, system efficacy, or broader market prevalence.", "cve": "" }, { "id": "LIMEN-WEB2-056", "title": "United States v. James Strahler II (first federal Take It Down Act conviction)", "summary": "A Columbus man pleaded guilty in what DOJ characterized as the first conviction under the criminal provision of the federal Take It Down Act for AI-generated abuse imagery.", "authority": "U.S. DOJ / S.D. Ohio", "anchor": "Guilty plea April 7, 2026 to charges including the Take It Down Act digital-forgeries provision", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "image generator (AI intimate imagery)", "source": "https://www.justice.gov/usao-sdoh/pr/columbus-man-pleads-guilty-cyberstalking-exes-creating-ai-generated-obscene-material", "caveat": "Very recent (April 2026); guilty plea with sentencing to follow; 'first conviction' is the DOJ characterization.", "cve": "" }, { "id": "LIMEN-WEB2-069", "title": "SEC v. Nathan Fuller & Privvy Investments (fake AI crypto trading bots)", "summary": "SEC alleges Fuller ran a $12.3M Ponzi promising 40-50% returns from non-existent AI trading bots, later using ChatGPT-generated KYC letters to stall withdrawals.", "authority": "SEC", "anchor": "SEC civil complaint filed May 28, 2026 (~$12.3M, 150+ investors)", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "purported AI-based crypto trading bots", "source": "https://www.investmentnews.com/regulation-legal-compliance/sec-sues-texas-man-over-alleged-123-million-ai-crypto-scheme/266827", "caveat": "Very recent complaint (May 2026), not adjudicated; confirmed via financial-trade press.", "cve": "" }, { "id": "LIMEN-WEB3-016", "title": "Tesla Autopilot fatal-crash civil verdict (Benavides v. Tesla, FL)", "summary": "A federal jury found Tesla's Autopilot design partly responsible for a fatal crash, the first US wrongful-death verdict tying liability to Autopilot operation.", "authority": "U.S. District Court, S.D. Florida (federal jury)", "anchor": "Verdict 1 Aug 2025 finding Tesla 33% liable, ~$243M; verdict upheld Feb 2026, on appeal", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.nbcnews.com/news/us-news/tesla-autopilot-crash-trial-verdict-partly-liable-rcna222344", "caveat": "Apportioned only 33% fault to Tesla (driver primarily at fault); Tesla is appealing, so liability is contested.", "cve": "" }, { "id": "LIMEN-WEB3-018", "title": "Tesla Full Self-Driving (FSD) reduced-visibility NHTSA investigation", "summary": "NHTSA is investigating Tesla FSD crashes in low-visibility conditions including a pedestrian death, citing camera-degradation detection/warning failures.", "authority": "NHTSA Office of Defects Investigation", "anchor": "Preliminary Evaluation opened 17 Oct 2024 (incl. 1 pedestrian fatality); upgraded to Engineering Analysis Mar 2026", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.nbcnews.com/tech/tech-news/tesla-faces-nhtsa-investigation-full-self-driving-fatal-collision-rcna176078", "caveat": "Open Engineering Analysis; no recall or final defect finding yet.", "cve": "" }, { "id": "LIMEN-WEB4-011", "title": "Spain fines age-verification provider Yoti over biometric processing", "summary": "AEPD fined Yoti EUR 950,000 for unlawful biometric processing, invalid consent and excessive retention in its age-verification app.", "authority": "AEPD", "anchor": "10 Mar 2026, EUR 950,000", "jurisdiction": "Spain", "country": "Spain", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "facial-scan biometric age/identity verification", "source": "https://www.dataguidance.com/news/spain-aepd-fines-yoti-eu950000-unlawful-processing", "caveat": "Yoti is appealing.", "cve": "" }, { "id": "LIMEN-WEB4-026", "title": "Robert Dillon wrongful arrest from 93% facial-recognition match (Florida)", "summary": "Police obtained a warrant on a 93% FRT match while allegedly concealing exculpatory evidence, jailing a man 300+ miles away.", "authority": "Jacksonville Beach PD (civil suit; ACLU)", "anchor": "Arrested Aug 2024; federal civil-rights suit filed June 2026", "jurisdiction": "United States (Florida)", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.aclu.org/press-releases/florida-man-sues-police-over-wrongful-arrest-due-to-false-facial-recognition-match", "caveat": "Newly filed complaint; untested.", "cve": "" }, { "id": "LIMEN-WEB4-030", "title": "Ford BlueCruise hands-free system — NTSB findings + NHTSA probe after fatal crashes", "summary": "Federal investigators concluded BlueCruise failed to brake before two fatal stationary-vehicle strikes, with driver over-reliance contributing.", "authority": "NTSB; NHTSA", "anchor": "NTSB board action 31 Mar 2026 on 2024 fatal crashes; NHTSA probe", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle (L2 hands-free)", "source": "https://www.detroitnews.com/story/business/autos/ford/2026/04/01/ntsb-ford-bluecruise-fatal-crashes-investigation-regulations-autonomous-vehicles/89417858007/", "caveat": "NTSB findings non-binding; NHTSA probe open, no recall/penalty.", "cve": "" }, { "id": "LIMEN-WEB4-048", "title": "Ibach v. Stewart — Alabama Supreme Court sanctions attorney $17,200 + filing restriction", "summary": "The Alabama Supreme Court dismissed the appeal, ordered $17,200, referred the attorney to the bar, and restricted his filings.", "authority": "Supreme Court of Alabama", "anchor": "2026; $17,200 + bar referral + solo-filing bar", "jurisdiction": "United States (Alabama Sup. Ct.)", "country": "United States", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://news.bloomberglaw.com/litigation/alabama-justices-toss-case-over-egregious-attorney-ai-misuse", "caveat": "Exact docket not confirmed.", "cve": "" }, { "id": "LIMEN-WEB4-067", "title": "eSafety Commissioner investigation into Grok-generated sexualised deepfakes on X", "summary": "Australia's online-safety regulator opened an investigation into Grok's generation of non-consensual sexualised deepfakes.", "authority": "eSafety Commissioner", "anchor": "Investigation announced late 2025/early 2026", "jurisdiction": "Australia", "country": "Australia", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.esafety.gov.au/newsroom/media-releases/esafety-raises-concerns-about-misuse-of-grok-to-generate-sexualised-content", "caveat": "No formal notice/fine/finding yet.", "cve": "" }, { "id": "LIMEN-WEB4-071", "title": "Ofcom formal investigation into X over Grok-generated deepfakes", "summary": "Ofcom opened a formal Online Safety Act investigation into X over Grok-generated deepfake intimate imagery and child sexual content.", "authority": "Ofcom (Online Safety Act 2023)", "anchor": "Investigation opened 12 Jan 2026", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.ofcom.org.uk/online-safety/illegal-and-harmful-content/investigation-into-x-and-scope-of-the-online-safety-act", "caveat": "No finding/penalty yet.", "cve": "" }, { "id": "LIMEN-WEB4-073", "title": "South Korea — 'Johnny Somali' convicted incl. distributing sexual deepfakes", "summary": "A South Korean court convicted the American streamer in part for distributing AI-generated sexual deepfakes.", "authority": "Seoul Western District Court", "anchor": "Convicted 15 Apr 2026; multiple counts", "jurisdiction": "South Korea", "country": "South Korea", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.japantimes.co.jp/life/2026/04/15/digital/johnny-somali-sentencing-south-korea/", "caveat": "Deepfake counts bundled with other charges.", "cve": "" }, { "id": "LIMEN-WEB4-079", "title": "Lancaster Country Day deepfake-nudes case — two students adjudicated delinquent", "summary": "Two Pennsylvania teens were adjudicated delinquent and given probation for AI deepfake nude images of classmates.", "authority": "Lancaster County Court of Common Pleas (juvenile)", "anchor": "Adjudicated; sentenced 25 Mar 2026; 59 felony counts", "jurisdiction": "United States (Pennsylvania)", "country": "United States", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://lancasteronline.com/news/local/2-former-lancaster-country-day-students-get-probation-for-making-nude-ai-images-of-peers/article_47ad9c75-75ea-4692-91c5-7e2ec7c9d753.html", "caveat": "Juvenile records sealed.", "cve": "" }, { "id": "LIMEN-WEB4-086", "title": "Brazil — MPF/ANPD order xAI to curb Grok sexualised deepfakes", "summary": "Brazilian regulators/prosecutors ordered xAI/X to prevent Grok generating non-consensual sexualised deepfakes, under threat of daily fines.", "authority": "Federal Public Ministry (MPF); ANPD", "anchor": "Recommendation 20 Jan 2026; escalated 11 Feb 2026", "jurisdiction": "Brazil", "country": "Brazil", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.techpolicy.press/x-tried-to-sidestep-brazils-inquiry-on-ai-deepfakes-the-government-just-pushed-back/", "caveat": "No final penalty imposed.", "cve": "" }, { "id": "LIMEN-WEB4-087", "title": "EU — European Commission opens DSA proceedings against X over Grok deepfakes", "summary": "The European Commission opened formal DSA proceedings against X over Grok-generated non-consensual intimate imagery and CSAM.", "authority": "European Commission (DSA)", "anchor": "Proceedings opened 26 Jan 2026 (IP/26/203)", "jurisdiction": "European Union", "country": "European Union", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://ec.europa.eu/commission/presscorner/detail/en/ip_26_203", "caveat": "Open proceedings; no finding/fine.", "cve": "" }, { "id": "LIMEN-WEB4-088", "title": "Netherlands — Amsterdam District Court orders X/xAI to stop Grok sexual-abuse content", "summary": "A Dutch court ordered X and xAI to stop Grok generating non-consensual sexual and child-abuse imagery, with a EUR 100,000/day penalty.", "authority": "Amsterdam District Court", "anchor": "Order 26 Mar 2026; EUR 100,000/day penalty", "jurisdiction": "Netherlands", "country": "Netherlands", "year": 2026, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.techpolicy.press/dutch-court-orders-x-grok-to-stop-aigenerated-sexual-abuse-content/", "caveat": "Civil injunction subject to possible appeal.", "cve": "" }, { "id": "LIMEN-WEB4-092", "title": "DOJ indicts former iLearningEngines CEO & CFO over fabricated AI-automation revenue", "summary": "DOJ alleges the executives inflated revenue by hundreds of millions via sham contracts while pitching iLearning as an AI automation company.", "authority": "DOJ (EDNY)", "anchor": "10-count indictment unsealed 17 Apr 2026", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI business-automation platform (Nasdaq: AILE)", "source": "https://www.justice.gov/usao-edny/pr/former-chief-executive-officer-and-chief-financial-officer-nasdaq-listed-company", "caveat": "Charges are accusations only; unproven.", "cve": "" }, { "id": "LIMEN-WEB4-095", "title": "FTC settlement vs Cox Media Group over fake 'Active Listening' AI ad-targeting", "summary": "FTC charged firms marketed an AI 'active listening' service that captured device voice for ad targeting, when it actually resold data-broker lists.", "authority": "FTC", "anchor": "Press release 21 May 2026; ~$930,000", "jurisdiction": "United States", "country": "United States", "year": 2026, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "claimed AI 'active listening' voice ad-targeting", "source": "https://www.ftc.gov/news-events/news/press-releases/2026/05/ftc-require-cox-media-group-two-other-firms-pay-nearly-1-million-settle-charges-they-deceived", "caveat": "Proposed settlement, not yet final; no admission.", "cve": "" }, { "id": "LIMEN-AUTH-000013", "title": "FTC Rytr complaint-final-order-set-aside chronology on AI-generated reviews, retained as a complex-posture negative result", "summary": "Official FTC complaint, 2024 final-order, and 2025 set-aside chronology for Rytr's AI review/testimonial generator, now fully document-backed on this host and usable as a bounded regulatory-reversibility / complex-posture negative result rather than a stable current-order exemplar.", "authority": "authoritative_candidate_hold_complex_posture", "anchor": "SRC-AUTH-0025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics", "Finance & insurance" ], "system": "generative text;review generation", "source": "", "caveat": "Use the full complaint/final-order/set-aside chronology together; do not reuse this row as a stable current final-order exemplar after the 2025 reopening and set-aside order.; This row is now strongest as a bounded official regulatory-reversibility / complex-posture negative result for review-manipulation governance, not as a denominator-widening authoritative-core promotion.; Do not convert the FTC's set-aside reasoning into a general scientific or legal vindication claim beyond the order text.", "cve": "" }, { "id": "LIMEN-WEB-006", "title": "Garcia v. Character Technologies: companion-chatbot product-liability suit over a teen's death", "summary": "Federal court (2025-05-21) denied dismissal in part, treating the chatbot as a product and rejecting a First-Amendment-speech defense; parties later settled (terms undisclosed).", "authority": "court_ruling_then_settled", "anchor": "M.D. Fla. 6:24-cv-01903, order 2025-05-21 (Conway J.)", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "companion chatbot", "source": "https://www.courtlistener.com/docket/69300919/garcia-v-character-technologies-inc/", "caveat": "Sensitive matter; cite only the docketed procedural rulings and the fact of settlement - no merits adjudication, no causal claim. | VERIFIED: interim MTD order (claims proceed), not a final liability finding; later settled.", "cve": "" }, { "id": "LIMEN-WEB-007", "title": "Mobley v. Workday: AI hiring-screening age-discrimination collective certified", "summary": "Court (2025-05-16) granted preliminary ADEA collective certification, allowing a disparate-impact claim that Workday's AI screening discriminated by age to proceed.", "authority": "court_certification", "anchor": "N.D. Cal. 3:23-cv-00770, order 2025-05-16 (Lin J.)", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "ai hiring screening", "source": "https://clearinghouse.net/case/44074/", "caveat": "Certification/procedural stage only; no liability finding on the merits. | VERIFIED: conditional ADEA collective certification, not a discrimination finding.", "cve": "" }, { "id": "LIMEN-WEB-010", "title": "Garante (Italy) EUR 5M fine on Luka Inc. over the Replika companion chatbot", "summary": "Garante (2025) fined Luka Inc. EUR 5M for processing without a legal basis and lacking age verification on the Replika emotional-AI chatbot; opened a separate training-data probe.", "authority": "regulator_final", "anchor": "EDPB national-news 2025 (Garante / Replika)", "jurisdiction": "Italy", "country": "Italy", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm", "Data protection" ], "system": "companion chatbot", "source": "https://www.edpb.europa.eu/news/national-news/2025/ai-italian-supervisory-authority-fines-company-behind-chatbot-replika_en", "caveat": "Fine may be subject to appeal; separate generative-training probe ongoing.", "cve": "" }, { "id": "LIMEN-WEB2-004", "title": "AEPD fines airport operator Aena and suspends facial-recognition boarding", "summary": "In Nov 2025 the AEPD fined Aena ~EUR 10M for an inadequate DPIA before enrolling ~40,000 travellers in facial-recognition boarding, and suspended the programme at eight airports.", "authority": "AEPD", "anchor": "AEPD decision Nov 2025; fine ~EUR 10M (Art. 35 GDPR), boarding suspended at 8 airports", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "airport biometric facial-recognition boarding", "source": "https://www.biometricupdate.com/202511/spanish-airport-operator-aena-fined-11m-over-biometric-boarding-program", "caveat": "Aena stated it will challenge the fine in court; not final. | VERIFIED: Aena has announced it will appeal; not final.", "cve": "" }, { "id": "LIMEN-WEB2-006", "title": "AEPD sanctions AI facial-recognition proctoring of online university exams", "summary": "The AEPD found AI facial-recognition proctoring of online exams to be unlawful biometric processing lacking a valid legal basis and sanctioned online universities including VIU.", "authority": "AEPD", "anchor": "AEPD sanctions of online universities (VIU), reported June 2025", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "AI exam-proctoring facial recognition", "source": "https://oecd.ai/en/incidents/2025-06-03-fe76", "caveat": "Specific decision references and exact fine amounts not located in the official record; sanctions may be appealed.", "cve": "" }, { "id": "LIMEN-WEB2-012", "title": "PIPC corrective order against DeepSeek over unlawful cross-border AI data transfers", "summary": "In April 2025 the PIPC found DeepSeek transferred Korean users' data to firms in China and the US without consent and ordered it to halt transfers and delete exported data after the app was pulled from Korean stores.", "authority": "PIPC", "anchor": "PIPC corrective order April 2025 (service suspended Feb 2025)", "jurisdiction": "South Korea", "country": "South Korea", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Data protection", "categories": [ "Data protection" ], "system": "generative-AI chatbot/app (DeepSeek)", "source": "https://www.cnbc.com/2025/04/24/south-korea-says-deepseek-transferred-user-data-to-china-us-without-consent.html", "caveat": "Corrective order rather than a confirmed monetary fine. | VERIFIED: corrective recommendation with no monetary fine.", "cve": "" }, { "id": "LIMEN-WEB2-015", "title": "In re IntelliVision Technologies Corp. (facial-recognition accuracy/bias claims)", "summary": "FTC finalized a 20-year consent order after alleging IntelliVision lacked support for claims of top accuracy, zero gender/racial bias, and spoof-proof performance.", "authority": "FTC", "anchor": "FTC Matter No. 232-3023; Docket C-4809; final order Jan 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "AI facial-recognition software", "source": "https://www.ftc.gov/legal-library/browse/cases-proceedings/232-3023-intellivision-matter", "caveat": "Administrative consent order; claims not litigated to a contested finding.", "cve": "" }, { "id": "LIMEN-WEB2-017", "title": "FTC v. Cleo AI, Inc. (AI cash-advance deception)", "summary": "FTC settled for $17M after alleging the AI-branded app deceived consumers about advance amounts/speed and used dark patterns to obstruct cancellation.", "authority": "FTC", "anchor": "S.D.N.Y. 1:25-cv-02594; $17M settlement Mar 27, 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "AI-branded fintech cash-advance app", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/03/cash-advance-company-cleo-ai-agrees-pay-17-million-result-ftc-lawsuit-charging-it-deceives-consumers", "caveat": "Settlement without admission; 'AI' is branding rather than the adjudicated harm mechanism.", "cve": "" }, { "id": "LIMEN-WEB2-018", "title": "FTC v. Ascend Ecom (Operation AI Comply, AI passive-income scheme)", "summary": "FTC obtained a court order banning Ascend Ecom's operators from business-opportunity marketing after alleging its 'AI-powered' passive-income tools defrauded consumers of at least $25M.", "authority": "FTC", "anchor": "C.D. Cal.; complaint Sep 25, 2024; ban order Jun 23, 2025; $25M judgment (partly suspended)", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-marketed e-commerce storefront automation", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/06/ftc-case-leads-order-banning-ascend-ecom-its-owners-business-opportunity-marketing", "caveat": "Monetary judgment largely suspended on inability-to-pay; harm figures are FTC allegations.", "cve": "" }, { "id": "LIMEN-WEB2-019", "title": "FTC v. Empire Holdings Group / Ecommerce Empire Builders (Operation AI Comply)", "summary": "FTC settled with a business-opportunity ban and a ~$9.79M judgment after alleging the scheme falsely promised an 'AI-powered Ecommerce Empire'.", "authority": "FTC", "anchor": "E.D. Pa. 2:24-cv-04949; settlement May 2025; ~$9.79M judgment (partly suspended)", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-marketed e-commerce business-opportunity", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/05/ftc-action-ends-ecommerce-empire-builders-online-business-opportunity-scam", "caveat": "Judgment partly suspended; allegations resolved by stipulated order.", "cve": "" }, { "id": "LIMEN-WEB2-021", "title": "FTC v. Click Profit, LLC et al. (Operation AI Comply)", "summary": "FTC obtained a TRO then a settlement permanently banning Click Profit's operators after alleging fabricated AI technology and fake brand partnerships cost consumers at least $14M.", "authority": "FTC", "anchor": "S.D. Fla. 1:25-cv-20973; settlement Aug 25, 2025; >$20M judgments", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-marketed e-commerce storefront automation", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/03/ftc-acts-stop-click-profit-online-business-opportunity-has-cost-consumers-least-14-million", "caveat": "Monetary judgments partly suspended; allegations resolved by stipulated order.", "cve": "" }, { "id": "LIMEN-WEB2-023", "title": "Texas AG investigation of Meta AI Studio and Character.AI (deceptive AI therapy chatbots)", "summary": "The Texas AG opened an investigation and issued CIDs alleging Meta and Character.AI deceptively marketed AI chatbots as mental-health/therapy tools to children.", "authority": "Texas state AG", "anchor": "Texas OAG press release Aug 18, 2025; Civil Investigative Demands", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "companion/persona chatbots presented as mental-health tools", "source": "https://www.texasattorneygeneral.gov/news/releases/attorney-general-ken-paxton-investigates-meta-and-characterai-misleading-children-deceptive-ai", "caveat": "Open investigation (CID stage); no complaint, settlement, or finding of liability. | VERIFIED: opened investigation (CIDs), not an enforcement finding.", "cve": "" }, { "id": "LIMEN-WEB2-025", "title": "Wadsworth v. Walmart — Morgan & Morgan attorneys fined for AI-hallucinated cases", "summary": "The court sanctioned three plaintiff attorneys (pro hac vice revoked + fines) for a motion citing eight nonexistent AI-generated cases.", "authority": "U.S. District Court for the District of Wyoming", "anchor": "No. 2:23-cv-00118-KHR (D. Wyo. Feb. 24, 2025)", "jurisdiction": "United States (D. Wyo.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "in-house generative AI tool", "source": "https://caselaw.findlaw.com/court/us-dis-crt-d-wyo/117003959.html", "caveat": "Sanctions mitigated by prompt withdrawal; the firm itself was not sanctioned.", "cve": "" }, { "id": "LIMEN-WEB2-026", "title": "Kohls v. Ellison — deepfake-law expert declaration excluded for AI-hallucinated citations", "summary": "The court excluded a Stanford professor's expert declaration defending a deepfake statute because it contained AI-generated nonexistent citations.", "authority": "U.S. District Court for the District of Minnesota", "anchor": "770 F. Supp. 3d 985 (D. Minn. Jan. 10, 2025); No. 24-cv-3754", "jurisdiction": "United States (D. Minn.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "GPT-4o (cited in expert declaration)", "source": "https://hlli.org/kohls-v-ellison/", "caveat": "The ruling went to admissibility, not the constitutional merits of the deepfake law.", "cve": "" }, { "id": "LIMEN-WEB2-027", "title": "Walters v. OpenAI — ChatGPT defamation suit dismissed on summary judgment", "summary": "The court granted OpenAI summary judgment on a defamation claim over a hallucinated output, holding no reasonable reader would treat it as fact.", "authority": "Superior Court of Gwinnett County, Georgia", "anchor": "No. 23-A-04860-2 (May 19, 2025)", "jurisdiction": "United States (Ga. Super. Ct.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "ChatGPT (LLM)", "source": "https://blog.ericgoldman.org/archives/2025/05/chatgpt-defeats-defamation-lawsuit-over-hallucination-walters-v-openai.htm", "caveat": "Single state trial-court order, not binding appellate precedent; turned on case-specific red flags.", "cve": "" }, { "id": "LIMEN-WEB2-030", "title": "Coomer v. Lindell — MyPillow attorneys sanctioned for ~30 AI-defective citations", "summary": "The court sanctioned two attorneys $3,000 each for a brief containing nearly 30 fabricated/defective AI-generated citations.", "authority": "U.S. District Court for the District of Colorado", "anchor": "No. 1:22-cv-01129 (D. Colo. July 7, 2025)", "jurisdiction": "United States (D. Colo.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "multiple LLMs (Copilot, Westlaw AI, ChatGPT)", "source": "https://reason.com/volokh/2025/07/07/6k-sanctions-for-apparent-ai-hallucinations-in-coomer-v-lindell-my-pillow-election-related-libel-suit/", "caveat": "First sanctions order; later 2026 orders involved separate repeat conduct. | VERIFIED: sanction is $6,000 total ($3,000 each).", "cve": "" }, { "id": "LIMEN-WEB2-031", "title": "Lacey v. State Farm — two law firms sanctioned $31,100 for AI-fabricated citations", "summary": "A Special Master imposed $31,100 in sanctions on two firms after roughly a third of citations in a brief were nonexistent or wrong, conduct deemed tantamount to bad faith.", "authority": "U.S. District Court for the Central District of California", "anchor": "No. 2:24-cv-05205 (C.D. Cal. May 6, 2025)", "jurisdiction": "United States (C.D. Cal.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "CoCounsel, Westlaw Precision, Google Gemini", "source": "https://incidentdatabase.ai/cite/1073/", "caveat": "Sanctions imposed by Special Master order within the case.", "cve": "" }, { "id": "LIMEN-WEB2-032", "title": "Thomson Reuters v. Ross Intelligence — fair-use defense rejected for AI training", "summary": "The court granted partial summary judgment for Thomson Reuters, holding ~2,243 Westlaw headnotes copyrightable and Ross's training use not fair use.", "authority": "U.S. District Court for the District of Delaware", "anchor": "765 F. Supp. 3d 382 (D. Del. Feb. 11, 2025); No. 1:20-cv-00613", "jurisdiction": "United States (D. Del.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "AI legal-research engine trained on Westlaw headnotes", "source": "https://www.dwt.com/blogs/artificial-intelligence-law-advisor/2025/02/reuters-ross-court-ruling-ai-copyright-fair-use", "caveat": "On interlocutory appeal to the Third Circuit; fair-use holding not yet final.", "cve": "" }, { "id": "LIMEN-WEB2-033", "title": "Bartz v. Anthropic — AI training on books held fair use; pirated copies not", "summary": "The court held on summary judgment that training Claude on lawfully acquired books was transformative fair use, but retaining pirated copies was infringing.", "authority": "U.S. District Court for the Northern District of California", "anchor": "No. 3:24-cv-05417-WHA (N.D. Cal. June 23, 2025)", "jurisdiction": "United States (N.D. Cal.)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "Claude (LLM trained on books)", "source": "https://www.akingump.com/en/insights/ai-law-and-regulation-tracker/district-court-rules-ai-training-can-be-fair-use-in-bartz-v-anthropic", "caveat": "Split ruling; piracy-liability set for trial and the matter later settled. | VERIFIED: split ruling - training fair use, pirated copies not; settled Sept 2025.", "cve": "" }, { "id": "LIMEN-WEB2-034", "title": "Kadrey v. Meta — AI training on books held fair use on this record", "summary": "The court granted Meta summary judgment on training-as-fair-use while cautioning the ruling reflected plaintiffs' weak record, not a general endorsement.", "authority": "U.S. District Court for the Northern District of California", "anchor": "No. 3:23-cv-03417-VC (N.D. Cal. June 25, 2025)", "jurisdiction": "United States (N.D. Cal.)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "LLaMA (LLM trained on books)", "source": "https://www.jw.com/news/insights-kadrey-meta-bartz-anthropic-ai-copyright/", "caveat": "Fact/record-specific; market-harm theory left open. | VERIFIED: record-bound win; court signaled stronger plaintiffs could prevail; distribution claims survived.", "cve": "" }, { "id": "LIMEN-WEB2-035", "title": "Concord Music Group v. Anthropic — lyrics preliminary injunction denied", "summary": "The court denied music publishers' preliminary injunction over Claude's training on lyrics after Anthropic stipulated to maintain output guardrails.", "authority": "U.S. District Court for the Northern District of California", "anchor": "No. 5:24-cv-03811-EKL (N.D. Cal. Mar. 25, 2025)", "jurisdiction": "United States (N.D. Cal.)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "Claude (LLM reproducing song lyrics)", "source": "https://docs.justia.com/cases/federal/district-courts/california/candce/5:2024cv03811/431519/321", "caveat": "Interim PI denial; underlying copyright merits unadjudicated.", "cve": "" }, { "id": "LIMEN-WEB2-037", "title": "Getty Images v. Stability AI — UK secondary-copyright claim dismissed", "summary": "The High Court rejected Getty's secondary-copyright-infringement claim and made only very limited trademark findings against early Stable Diffusion versions.", "authority": "High Court of Justice (Chancery Division)", "anchor": "[2025] EWHC 2863 (Ch) (4 Nov. 2025)", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "Stable Diffusion (image generator)", "source": "https://www.twobirds.com/en/insights/2025/uk/stability-ai-defeats-getty-images-copyright-claims-in-first-of-its-kind-dispute-before-the-high-cour", "caveat": "Getty's primary training claims were narrowed before judgment; permission to appeal granted.", "cve": "" }, { "id": "LIMEN-WEB2-039", "title": "Ayinde v. London Borough of Haringey — High Court warns lawyers over fake AI citations", "summary": "In a Hamid-jurisdiction judgment the Divisional Court found legal teams had advanced fabricated AI-generated authorities and warned of severe sanction, referring matters to regulators.", "authority": "High Court of Justice (Divisional Court)", "anchor": "[2025] EWHC 1383 (Admin) (6 June 2025)", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative AI (LLM)", "source": "https://www.judiciary.uk/wp-content/uploads/2025/06/Ayinde-v-London-Borough-of-Haringey-and-Al-Haroun-v-Qatar-National-Bank.pdf", "caveat": "Court declined contempt against a junior barrister; sanctions left to professional regulators.", "cve": "" }, { "id": "LIMEN-WEB2-041", "title": "Randal Reid wrongful arrest (Jefferson Parish Clearview-based misidentification)", "summary": "Reid was jailed six days in 2022 on a Louisiana warrant generated by a facial-recognition match while he had never been to the state, resolved by a $200,000 settlement in 2025.", "authority": "U.S. District Court, E.D. Louisiana", "anchor": "$200,000 settlement accepted May 13, 2025 (sealed)", "jurisdiction": "United States (Louisiana)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.nola.com/news/jefferson_parish/facial-recognition-sheriff/article_3ed262bc-e360-4383-82fb-728b3783cb93.html", "caveat": "No public docket number; settlement sealed; amount rests on authoritative local reporting.", "cve": "" }, { "id": "LIMEN-WEB2-044", "title": "Porcha Woodruff wrongful arrest (Detroit facial recognition; suit dismissed)", "summary": "Woodruff, eight months pregnant, was wrongfully arrested in 2023 after a facial-recognition match seeded a photo lineup; her federal suit was dismissed in 2025 on probable-cause grounds.", "authority": "U.S. District Court, E.D. Michigan", "anchor": "Woodruff v. Oliver; suit dismissed by order dated Aug. 5, 2025; appeal planned", "jurisdiction": "United States (Michigan)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.cbsnews.com/detroit/news/woman-wrongly-accused-carjacking-loses-lawsuit-detroit-police-used-facial-tech/", "caveat": "Court dismissed the claim; appeal planned, so liability not established.", "cve": "" }, { "id": "LIMEN-WEB2-050", "title": "Estate of Lokken v. UnitedHealth Group (nH Predict AI post-acute-care denials)", "summary": "A live federal class action alleging UnitedHealth/naviHealth used the nH Predict algorithm to systematically deny Medicare Advantage post-acute care, with two common-law claims surviving dismissal.", "authority": "U.S. District Court for the District of Minnesota", "anchor": "No. 0:23-cv-03514; filed Nov 2023; MTD ruling 13 Feb 2025 let breach-of-contract and good-faith claims proceed", "jurisdiction": "United States (Minnesota)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "nH Predict predictive algorithm for Medicare Advantage post-acute coverage cutoffs", "source": "https://www.classaction.org/media/the-estate-of-gene-b-lokken-et-al-v-unitedhealth-group-inc-et-al.pdf", "caveat": "Pleading/discovery stage; no liability finding; the alleged ~90% appeal-reversal rate is a plaintiff allegation; five of seven counts were dismissed.", "cve": "" }, { "id": "LIMEN-WEB2-052", "title": "Kisting-Leung v. Cigna (PxDx automated claim-denial algorithm)", "summary": "A live federal class action alleging Cigna's PxDx system enabled mass claim denials without individualized medical review, with ERISA and UCL claims surviving dismissal.", "authority": "U.S. District Court for the Eastern District of California", "anchor": "No. 2:23-cv-01477; filed 24 Jul 2023; MTD order 31 Mar 2025 allowed ERISA and California UCL claims to proceed", "jurisdiction": "United States (California)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "PxDx automated procedure-to-diagnosis claim-review system", "source": "https://docs.justia.com/cases/federal/district-courts/california/caedce/2:2023cv01477/431351/55", "caveat": "Interim stage; some named plaintiffs found to lack standing; the 300,000-denials/1.2-second figures derive from ProPublica reporting, not a court finding. | VERIFIED: MTD order dated 30 Mar 2025 (anchor said 31).", "cve": "" }, { "id": "LIMEN-WEB2-054", "title": "Dazhon Darien convicted for AI voice-clone framing of Pikesville HS principal", "summary": "A former athletic director was prosecuted and convicted for using an AI voice clone to fabricate racist remarks attributed to the principal.", "authority": "Baltimore County Circuit Court (Maryland)", "anchor": "Charged April 25, 2024; pleaded guilty and sentenced (4 months) in 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "synthetic voice", "source": "https://www.cbsnews.com/baltimore/news/maryland-framed-principal-racist-ai-generated-voice/", "caveat": "Conviction via guilty plea to a reduced charge (disrupting school operations).", "cve": "" }, { "id": "LIMEN-WEB2-060", "title": "NC State sorority AI deepfake-porn case: Hayne Beard charged", "summary": "A man was criminally charged in North Carolina for creating and posting AI-manipulated explicit images of NC State sorority members.", "authority": "Wake County District Attorney (North Carolina)", "anchor": "Charged with two counts of misdemeanor stalking; arrest ~April 19, 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.wral.com/news/local/police-charge-man-ai-porn-nc-students-2025/", "caveat": "Charging-stage misdemeanor stalking; not yet adjudicated.", "cve": "" }, { "id": "LIMEN-WEB2-063", "title": "SEC v. Presto Automation Inc. (first AI-washing action vs a public company)", "summary": "SEC's first AI-washing action against a public company; settled cease-and-desist finding Presto misrepresented that proprietary AI eliminated human order-taking.", "authority": "SEC", "anchor": "Securities Act Rel. 11352 / Exchange Act Rel. 102177, Admin. Proc. 3-22413, Jan 14, 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "voice/speech-recognition AI for drive-thru order taking", "source": "https://www.sec.gov/files/litigation/admin/2025/33-11352.pdf", "caveat": "Settled without admitting/denying findings.", "cve": "" }, { "id": "LIMEN-WEB2-068", "title": "SEC & DOJ v. Alexander Beckman (GameOn AI chatbot startup fraud)", "summary": "SEC/DOJ allege Beckman defrauded investors of $60M+ at the AI startup GameOn via forged bank statements, a fake audit and fabricated revenue.", "authority": "SEC + DOJ N.D. Cal.", "anchor": "SEC Litigation Release LR-26232, Jan 23, 2025; DOJ indictment 25-CR-00012", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI sports/fan-engagement chatbot platform", "source": "https://www.sec.gov/enforcement-litigation/litigation-releases/lr-26232", "caveat": "Charged conduct centers on financial forgery rather than AI falsity; allegations unadjudicated. | VERIFIED: SEC release dated 24 Jan 2025; DOJ indictment unsealed 21 Jan 2025.", "cve": "" }, { "id": "LIMEN-WEB2-071", "title": "ACLU charge v. Intuit & HireVue (AI video-interview disability/race discrimination)", "summary": "ACLU alleges HireVue's AI video interview, lacking captioning, discriminated against a deaf Indigenous employee under the ADA, Title VII and Colorado law.", "authority": "EEOC + Colorado Civil Rights Division", "anchor": "ACLU charge filed Mar 19, 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "HireVue automated video-interview / speech-recognition assessment", "source": "https://www.hrdive.com/news/ai-intuit-hirevue-deaf-indigenous-employee-discrimination-aclu/743273/", "caveat": "Administrative charge, not adjudicated; Intuit denies and HireVue says no AI-backed assessment was used.", "cve": "" }, { "id": "LIMEN-WEB2-072", "title": "Harper v. Sirius XM Radio (AI resume-screening race discrimination class action)", "summary": "Plaintiff alleges Sirius XM's AI hiring tool used race-proxy data points causing disparate treatment/impact after he was rejected for ~149 of 150 applications.", "authority": "U.S. District Court, E.D. Michigan", "anchor": "No. 2:25-cv-12403 (E.D. Mich.), filed Aug 4, 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "AI-assisted applicant tracking / resume-screening", "source": "https://www.workforcebulletin.com/artificial-intelligence-bias-harper-v-sirius-xm-challenges-algorithmic-discrimination-in-hiring", "caveat": "Recently filed complaint; allegations unadjudicated and class status not certified.", "cve": "" }, { "id": "LIMEN-WEB3-013", "title": "Spain BOSCO energy-subsidy eligibility algorithm transparency case", "summary": "After BOSCO wrongly denied energy-bill subsidies, Spain's Supreme Court ruled in 2025 that the government must disclose the algorithm's source code, a public-algorithm transparency precedent.", "authority": "Tribunal Supremo; Audiencia Nacional", "anchor": "Supreme Court ruling Sept 2025 ordering source-code release, reversing Audiencia Nacional 30 Apr 2024", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "eligibility-determination algorithm for the electricity social bonus", "source": "https://civio.es/novedades/2025/09/18/civio-pulls-back-the-curtain-on-public-algorithms-spains-supreme-court-orders-the-government-to-release-boscos-source-code/", "caveat": "Concerns transparency/source-code access, not a finding that BOSCO's outputs were unlawful; exact Sept 2025 day varies across reports.", "cve": "" }, { "id": "LIMEN-WEB3-020", "title": "Waymo 5th-gen ADS barrier-collision recall", "summary": "Waymo recalled software after its driverless vehicles repeatedly collided with stationary/semi-stationary roadway barriers under an open NHTSA investigation.", "authority": "NHTSA", "anchor": "Recall 25E-034 (May 2025), 1,212 vehicles; followed a 2024 preliminary evaluation", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://static.nhtsa.gov/odi/rcl/2025/RCLRPT-25E034-2471.PDF", "caveat": "No injuries reported; a software fix was deployed before the formal recall (property-damage/defect case).", "cve": "" }, { "id": "LIMEN-WEB3-035", "title": "ASIC Australia: former Metigy CEO pleaded guilty over false 'AI' company claims", "summary": "The CEO of AI-marketing startup Metigy pleaded guilty to making false/misleading statements to investors (revenue ~AUD 43,000 vs claims backing ~AUD 39M raised).", "authority": "ASIC / Commonwealth DPP", "anchor": "Charges 18 Nov 2024 (24-248MR); guilty plea Sept 2025 (25-263MR)", "jurisdiction": "Australia", "country": "Australia", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "purported AI-driven marketing software platform", "source": "https://www.asic.gov.au/about-asic/news-centre/find-a-media-release/2025-releases/25-263mr-former-ceo-of-ai-marketing-company-metigy-pleads-guilty-to-misleading-investors-and-dishonestly-using-his-position/", "caveat": "Guilty plea to misleading-statements/dishonesty broadly; sentencing pending and charges centre on financials. | VERIFIED: charge/first appearance 8 Nov 2024; guilty plea announced 4 Nov 2025 (anchor dates off).", "cve": "" }, { "id": "LIMEN-WEB3-038", "title": "Kenya: government suspension and High Court ruling against Worldcoin iris collection", "summary": "After suspending Worldcoin in 2023, the Kenya High Court ruled its iris-biometric collection unlawful under Kenyan data-protection law and ordered deletion of all collected biometric data.", "authority": "Government of Kenya / High Court of Kenya", "anchor": "Govt suspension Aug 2023; High Court ruling 5 May 2025", "jurisdiction": "Kenya", "country": "Kenya", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "iris-scanning biometric proof-of-personhood", "source": "https://tech-ish.com/2025/05/05/worldcoin-high-court-ruling-kenya/", "caveat": "2023 step was an executive suspension; binding illegality finding came via the 2025 court ruling.", "cve": "" }, { "id": "LIMEN-WEB3-039", "title": "ANPD Brazil: suspension of Worldcoin biometric-for-crypto collection", "summary": "ANPD ordered Tools for Humanity to stop offering crypto compensation for iris/biometric data, holding financial incentives could vitiate free, informed consent, with daily fines for non-compliance.", "authority": "Autoridade Nacional de Protecao de Dados (ANPD)", "anchor": "Preventive suspension Jan 2025; appeal rejected Feb 2025", "jurisdiction": "Brazil", "country": "Brazil", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Data protection", "categories": [ "Data protection" ], "system": "iris-scanning biometric proof-of-personhood", "source": "https://www.biometricupdate.com/202501/brazil-bans-tfh-from-offering-compensation-for-biometric-data", "caveat": "Preventive/administrative measure (upheld on appeal) rather than a final adjudicated penalty.", "cve": "" }, { "id": "LIMEN-WEB4-007", "title": "Ireland fines Dept. of Social Protection over Public Services Card facial recognition", "summary": "The DPC fined the DSP EUR 550,000 for facial-recognition biometric matching in Public Services Card registration without a valid lawful basis.", "authority": "Data Protection Commission", "anchor": "12 June 2025, EUR 550,000 + reprimand", "jurisdiction": "Ireland", "country": "Ireland", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-template biometric matching for identity de-duplication", "source": "https://www.rte.ie/news/business/2025/0612/1518065-data-watchdog-issues-550k-fine-over-public-services-card/", "caveat": "DSP is challenging the fine.", "cve": "" }, { "id": "LIMEN-WEB4-008", "title": "Spain fines LaLiga for stadium biometric access control", "summary": "AEPD fined LaLiga EUR 1M for mandating biometric stadium access without a compliant DPIA.", "authority": "AEPD", "anchor": "EXP202315637, EUR 1,000,000 (contested Mar 2025)", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "biometric stadium access control", "source": "https://www.biometricupdate.com/202503/laliga-disputes-e1m-fine-from-aepd-over-stadium-biometrics", "caveat": "LaLiga has appealed.", "cve": "" }, { "id": "LIMEN-WEB4-014", "title": "Germany (Hamburg) fines a financial company over automated credit decisions", "summary": "The Hamburg DPA fined a financial company EUR 492,000 for automated credit-card rejection decisions without adequate explanation.", "authority": "Hamburg DPA (HmbBfDI)", "anchor": "2025, EUR 492,000", "jurisdiction": "Germany", "country": "Germany", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "solely-automated credit-card decisioning (Art. 22)", "source": "https://www.clydeco.com/en/insights/2025/10/lessons-from-hamburg-commissioner-for-data-protect", "caveat": "Company unnamed.", "cve": "" }, { "id": "LIMEN-WEB4-017", "title": "Italy fines a high school over fingerprint attendance for staff", "summary": "The Garante fined a high school EUR 4,000 for using staff fingerprints for attendance, holding consent invalid in employment.", "authority": "Garante", "anchor": "27 Mar 2025, EUR 4,000", "jurisdiction": "Italy", "country": "Italy", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "fingerprint biometric staff attendance", "source": "https://www.edpb.europa.eu/news/national-news/2025/biometrics-attendance-recording-italian-sa-fines-high-school_en", "caveat": "None material; EDPB-confirmed.", "cve": "" }, { "id": "LIMEN-WEB4-023", "title": "Norway fines Grindr over sharing sensitive profile data with adtech", "summary": "Datatilsynet fined Grindr NOK 65M for unlawfully sharing users' data (revealing likely sexual orientation) with ad partners; upheld 2025.", "authority": "Datatilsynet", "anchor": "15 Dec 2021, NOK 65M; upheld on appeal 2025", "jurisdiction": "Norway", "country": "Norway", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "adtech profiling/data-sharing pipeline", "source": "https://www.datatilsynet.no/en/news/news-2025/the-court-of-appeal-upholds-the-fine-against-grindr/", "caveat": "None material; regulator-direct.", "cve": "" }, { "id": "LIMEN-WEB4-024", "title": "Alonzo Sawyer wrongful arrest via facial-recognition match (Maryland)", "summary": "An innocent man was jailed on a false FRT match before the true suspect was identified.", "authority": "Maryland Transit Administration Police / state court", "anchor": "Arrested spring 2022; charges dropped; ACLU 2025 tally", "jurisdiction": "United States (Maryland)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.aclu.org/news/privacy-technology/more-than-a-dozen-wrongful-arrests-due-to-police-reliance-on-facial-recognition-technology", "caveat": "No civil judgment/settlement on record; advocacy-group documentation.", "cve": "" }, { "id": "LIMEN-WEB4-025", "title": "Trevis Williams wrongful arrest after NYPD facial-recognition match", "summary": "NYPD arrested a man markedly different from the suspect, with phone data placing him 12 miles away, on a flawed FRT lead.", "authority": "NYPD / NY County DA (charges dismissed)", "anchor": "Arrested Apr 2025; dismissed July 2025", "jurisdiction": "United States (New York)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://abc7ny.com/post/man-falsely-jailed-nypds-facial-recognition-surveillance-tech-failed/17664671/", "caveat": "Prosecutorial dismissal; no civil judgment yet.", "cve": "" }, { "id": "LIMEN-WEB4-031", "title": "Zoox robotaxi hard-braking software recall", "summary": "Amazon's Zoox recalled its driverless fleet after the autonomy stack mispredicted collisions and braked unsafely hard.", "authority": "NHTSA", "anchor": "Recall 25E037, 19 Mar 2025; 258 vehicles", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://static.nhtsa.gov/odi/rcl/2025/RCLRPT-25E037-4912.pdf", "caveat": "Voluntary recall; manufacturer root-cause framing. | VERIFIED: Correct the recall to NHTSA campaign 25E019 (report received 14 Mar 2025) for the 258-vehicle hard-braking software recall; 25E037 is a separate May 2025 Zoox recall.", "cve": "" }, { "id": "LIMEN-WEB4-032", "title": "Zoox robotaxi lane-crossing software recall", "summary": "Zoox's third recall in eight months addressed autonomy that crossed into oncoming lanes near intersections.", "authority": "NHTSA", "anchor": "Recall Dec 2025; 332 robotaxis", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://techcrunch.com/2025/12/23/zoox-issues-software-recall-over-lane-crossings/", "caveat": "Recall from manufacturer filing/trade reporting; injury record not established.", "cve": "" }, { "id": "LIMEN-WEB4-033", "title": "Waymo robotaxi NHTSA investigation for passing a stopped school bus", "summary": "A federal defects probe opened after a driverless Waymo passed a stopped school bus with its stop-arm deployed.", "authority": "NHTSA Office of Defects Investigation", "anchor": "ODI investigation opened Oct 2025 (~2,000 vehicles)", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.cbsnews.com/news/waymo-investigation-nhtsa-robotaxis-passing-school-bus/", "caveat": "Open investigation; no defect finding/recall (distinct from Waymo's barrier recall).", "cve": "" }, { "id": "LIMEN-WEB4-036", "title": "Mid Central Operating Engineers v. HoosierVac — attorney sanctioned + bar referral for AI cases", "summary": "A magistrate sanctioned the attorney $6,000 and referred him for discipline over AI-fabricated citations.", "authority": "U.S. District Court, S.D. Ind.", "anchor": "No. 2:24-cv-00326; 2025 WL 1511211 (May 28 2025); $6,000 + referral", "jurisdiction": "United States (S.D. Ind.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.casemine.com/judgement/us/68391a2c20bd54916e6efca0", "caveat": "Bar-referral outcome unknown. | VERIFIED: Magistrate Judge Dinsmore's order; he had recommended $15,000 but the final imposed sanction was $6,000 plus bar awareness.", "cve": "" }, { "id": "LIMEN-WEB4-037", "title": "Ko v. Li — Ontario family-law counsel show-cause for contempt over AI citations", "summary": "An Ontario court ordered counsel to show cause for contempt over AI-hallucinated citations, then accepted her apology.", "authority": "Ontario Superior Court of Justice", "anchor": "2025 ONSC 2766 / 2025 ONSC 2965 (Myers J.)", "jurisdiction": "Canada (Ontario Superior Court)", "country": "Canada", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.canadianlawyermag.com/practice-areas/family/superior-court-says-family-law-case-factum-may-have-fake-legal-citations-due-to-ai-hallucinations/392433", "caveat": "No monetary sanction ultimately imposed; remedial outcome.", "cve": "" }, { "id": "LIMEN-WEB4-039", "title": "Nguyen v. Wheeler — counsel sanctioned $1,000 for four nonexistent AI cases", "summary": "The court imposed a $1,000 sanction for AI-fabricated authority.", "authority": "U.S. District Court, E.D. Ark.", "anchor": "No. 4:24-cv-00815 (Mar 2025); $1,000", "jurisdiction": "United States (E.D. Ark.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Order date from secondary review.", "cve": "" }, { "id": "LIMEN-WEB4-040", "title": "Benjamin v. Costco — counsel sanctioned for five fabricated/misattributed AI citations", "summary": "The court sanctioned counsel for five fabricated/misattributed citations, reduced for candor.", "authority": "U.S. District Court, E.D.N.Y.", "anchor": "No. 1:24-cv-07399 (Apr 24 2025)", "jurisdiction": "United States (E.D.N.Y.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.esquiresolutions.com/the-sirens-song-of-generative-ai-in-pleadings/", "caveat": "Exact dollar figure not pinned to the order. | VERIFIED: Docket number is 2:24-cv-07399 (not 1:24-cv-07399); Magistrate Judge Dunst imposed a $1,000 sanction plus client notice and AI CLE.", "cve": "" }, { "id": "LIMEN-WEB4-041", "title": "Bevins v. Colgate-Palmolive — attorney sanctioned over fabricated AI citations", "summary": "A federal court found fabricated citations and moved to sanction counsel.", "authority": "U.S. District Court, E.D. Pa.", "anchor": "E.D. Pa. 2025; sanctions imposed", "jurisdiction": "United States (E.D. Pa.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.legal.io/articles/5609086/Fake-Case-Citations-Land-Two-Attorneys-in-Hot-Water-Over-AI-Misuse", "caveat": "Exact docket/sanction amount not confirmed. | VERIFIED: Sanction was non-monetary (appearance struck, bar-regulator notice, client notification) per Judge Kai Scott, No. 25-576, E.D. Pa., Apr 10 2025; not framed as a fine.", "cve": "" }, { "id": "LIMEN-WEB4-042", "title": "Dehghani v. Castro — counsel referred to NM and TX bars for nonexistent AI cases", "summary": "The court referred counsel to two state bars for AI-fabricated citations.", "authority": "U.S. District Court, D.N.M.", "anchor": "No. 2:25-cv-00052 (Apr 2 2025); bar referrals", "jurisdiction": "United States (D.N.M.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://law.justia.com/cases/federal/district-courts/new-mexico/nmdce/2:2025cv00052/511942/35/", "caveat": "Bar-referral outcomes pending.", "cve": "" }, { "id": "LIMEN-WEB4-043", "title": "In re Loletha Hale — five-year disclosure sanction for AI-fabricated citations", "summary": "The court imposed a five-year disclosure/notification sanction for AI-fabricated citations.", "authority": "U.S. District Court, N.D. Ga.", "anchor": "Oct 28 2025; client-notification + 5-year disclosure", "jurisdiction": "United States (N.D. Ga.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Caption summarized from a review article.", "cve": "" }, { "id": "LIMEN-WEB4-044", "title": "Dubinin v. Papazian — case dismissed + fees + bar referrals over AI citations", "summary": "The court dismissed the case, awarded fees, and referred counsel to state bars for AI-fabricated citations.", "authority": "U.S. District Court, S.D. Fla.", "anchor": "Nov 21 2025; dismissal + fees + referrals", "jurisdiction": "United States (S.D. Fla.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Docket/fee amount from a review article.", "cve": "" }, { "id": "LIMEN-WEB4-045", "title": "Tercero v. Sacramento Logistics — counsel fined $1,500 + CA bar referral", "summary": "The court fined counsel $1,500 and referred him to the California bar.", "authority": "U.S. District Court, E.D. Cal.", "anchor": "Sept 9 2025; $1,500 + referral", "jurisdiction": "United States (E.D. Cal.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Docket from a review article.", "cve": "" }, { "id": "LIMEN-WEB4-046", "title": "Idehen v. Stoute-Phillip — show-cause over 'egregious' AI-fabricated citations", "summary": "A New York civil court issued a show-cause order over egregious AI-fabricated citations.", "authority": "N.Y. Civil Court, Queens County", "anchor": "July 29 2025; show-cause", "jurisdiction": "United States (NY Civil Court, Queens)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Final sanction not confirmed.", "cve": "" }, { "id": "LIMEN-WEB4-047", "title": "Johnson v. Dunn — three Butler Snow attorneys disqualified + bar referral over AI citations", "summary": "The court disqualified three attorneys and referred them to the state bar for AI-fabricated citations.", "authority": "U.S. District Court, N.D. Ala.", "anchor": "792 F.Supp.3d 1241 (2025); order July 23 2025", "jurisdiction": "United States (N.D. Ala.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.courthousenews.com/johnson-vs-dunn-attorney-sanctions-order/", "caveat": "Bar referral pending.", "cve": "" }, { "id": "LIMEN-WEB4-049", "title": "Puerto Rico Soccer League v. FPF — counsel ordered ~$24,000 fees for 55+ defective AI citations", "summary": "The court ordered ~$24,000 in fees over dozens of defective/nonexistent citations.", "authority": "U.S. District Court, D.P.R.", "anchor": "2025; ~$24,000 fees", "jurisdiction": "United States (D.P.R.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://reason.com/volokh/2025/09/24/over-24k-in-sanctions-for-alleged-ai-hallucinations-in-case-against-fifa/", "caveat": "Counsel denied using AI; court inferred it.", "cve": "" }, { "id": "LIMEN-WEB4-050", "title": "Crespo v. Tesla — pro se plaintiff ordered to pay Tesla fees for AI-hallucinated cases", "summary": "The court ordered the plaintiff to pay Tesla's fees for filings citing AI-hallucinated cases.", "authority": "U.S. District Court, S.D. Fla.", "anchor": "2025 WL 1921903 (July 11 2025)", "jurisdiction": "United States (S.D. Fla.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "http://evan.law/2025/07/15/tesla-awarded-sanctions-where-opposing-party-cited-ai-generated-cases-in-discovery-briefing/", "caveat": "Pro se; amount from secondary source.", "cve": "" }, { "id": "LIMEN-WEB4-051", "title": "Bunce v. Visual Technology Innovations — attorney sanctioned $16,270 for ChatGPT cases", "summary": "The court sanctioned the attorney $16,270.47 for AI-hallucinated citations.", "authority": "U.S. District Court, E.D. Pa.", "anchor": "2025 WL 662398 (Feb 27 2025); $16,270.47", "jurisdiction": "United States (E.D. Pa.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.casemine.com/judgement/us/67d7b935ba1fd11f94020e1a", "caveat": "None material.", "cve": "" }, { "id": "LIMEN-WEB4-052", "title": "Versant Funding v. Teras Breakbulk — two attorneys sanctioned for hallucinated AI citation", "summary": "The court sanctioned two attorneys for filing a hallucinated AI citation without verification.", "authority": "U.S. District Court, S.D. Fla.", "anchor": "May 20 2025 (Matthewman M.J.)", "jurisdiction": "United States (S.D. Fla.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://edrm.net/2025/08/reasonable-or-overreach-rethinking-sanctions-for-ai-hallucinations-in-legal-filings/", "caveat": "Amount/docket not confirmed from primary order.", "cve": "" }, { "id": "LIMEN-WEB4-053", "title": "Richburg v. Glyndon Square — bankruptcy counsel sanctioned with AI-ethics CLE", "summary": "The bankruptcy court sanctioned counsel with mandatory ethics/AI CLE for fabricated citations.", "authority": "U.S. Bankruptcy Court, D.S.C.", "anchor": "Adv. 25-80037-EG (Aug 27 2025); CLE", "jurisdiction": "United States (Bankr. D.S.C.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://naturalandartificiallaw.com/ai-hallucinations-and-or-fabricated-citations-2/", "caveat": "From a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-054", "title": "Allbaugh v. University of Scranton — pro se plaintiff fined $1,000 for AI citations", "summary": "The court fined the pro se plaintiff $1,000 for AI-fabricated citations.", "authority": "U.S. District Court, M.D. Pa.", "anchor": "No. 3:24-cv-02237 (Aug 28 2025); $1,000", "jurisdiction": "United States (M.D. Pa.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://naturalandartificiallaw.com/ai-hallucinations-and-or-fabricated-citations-2/", "caveat": "From a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-055", "title": "13th Judicial Circuit Clerk v. Rangel — counsel referred to Florida Bar over AI citations", "summary": "A Florida appellate court referred counsel to the Florida Bar over AI-fabricated citations.", "authority": "Florida Second District Court of Appeal", "anchor": "No. 2D2024-1772 (Aug 29 2025); bar referral", "jurisdiction": "United States (Fla. 2d DCA)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://naturalandartificiallaw.com/ai-hallucinations-and-or-fabricated-citations-2/", "caveat": "From a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-056", "title": "Multiphone Latin America v. Millicom — two attorneys referred over AI citations", "summary": "The court referred two attorneys for investigation over AI-fabricated citations.", "authority": "U.S. District Court, S.D. Fla.", "anchor": "No. 25-23249-CIV (Aug 28 2025); referrals", "jurisdiction": "United States (S.D. Fla.)", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://naturalandartificiallaw.com/ai-hallucinations-and-or-fabricated-citations-2/", "caveat": "From a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-057", "title": "Thackston v. Driscoll — Rule 11 sanctions recommended for AI citations", "summary": "A magistrate recommended Rule 11 sanctions and CLE for AI-fabricated citations.", "authority": "U.S. District Court, W.D. Tex.", "anchor": "No. SA-24-CV-00276 (Aug 2025); recommendation", "jurisdiction": "United States (W.D. Tex.)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://naturalandartificiallaw.com/ai-hallucinations-and-or-fabricated-citations-2/", "caveat": "Pending district-judge adoption.", "cve": "" }, { "id": "LIMEN-WEB4-058", "title": "Valu v. Minister for Immigration (Dayal) — first Australian lawyer penalised for AI citations", "summary": "An Australian court referred the solicitor to the NSW Legal Services Commissioner for AI-fabricated citations; he was later barred from practising as a principal.", "authority": "Federal Circuit and Family Court of Australia (Div. 2)", "anchor": "[2025] FedCFamC2G 95 (31 Jan 2025); referred to NSW LSC", "jurisdiction": "Australia (Fed. Circuit & Family Court)", "country": "Australia", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://ia.acs.org.au/article/2025/first-australian-lawyer-penalised-for-ai-blunder.html", "caveat": "Principal-practice restriction from news reporting. | VERIFIED: Valu v. Minister for Immigration (No 2) [2025] FedCFamC2G 95 (31 Jan 2025): the practitioner was referred to the NSW Legal Services Commissioner; the 'first Australian lawyer penalised' descriptor refers to the separate Dayal matter (Victoria).", "cve": "" }, { "id": "LIMEN-WEB4-059", "title": "Re Walker — Supreme Court of Victoria reprimands solicitor for AI authorities", "summary": "The Supreme Court of Victoria reprimanded the solicitor for unprofessional conduct over unverified AI authorities.", "authority": "Supreme Court of Victoria", "anchor": "[2025] VSC 714 (24 Nov 2025)", "jurisdiction": "Australia (Supreme Court of Victoria)", "country": "Australia", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://kirstenegroth.blog/ai-hallucination-cases/", "caveat": "Citation from a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-060", "title": "Murray v. State of Victoria (Wamba Wemba) — solicitor ordered to pay indemnity costs for AI use", "summary": "The Federal Court ordered the solicitor to personally bear indemnity costs for unverified AI in prepared documents.", "authority": "Federal Court of Australia", "anchor": "[2025] FCA 731 (22 Apr 2025)", "jurisdiction": "Australia (Federal Court)", "country": "Australia", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://kirstenegroth.blog/ai-hallucination-cases/", "caveat": "Citation from a tracker.", "cve": "" }, { "id": "LIMEN-WEB4-061", "title": "The Father v The Mother — UK High Court costs + referral over faked AI cases", "summary": "The High Court ordered a GBP 5,900.63 costs sanction and referred the matter to a Presiding Judge over faked AI cases.", "authority": "High Court of Justice (Family Division)", "anchor": "[2025] EWHC 2135 (Fam) (30 July 2025); GBP 5,900.63 costs", "jurisdiction": "United Kingdom (EWHC Family)", "country": "United Kingdom", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.bailii.org/ew/cases/EWHC/Fam/2025/2135.html", "caveat": "Litigant-in-person; no bar discipline.", "cve": "" }, { "id": "LIMEN-WEB4-062", "title": "Hussein v. Canada — Federal Court special costs for fabricated AI citations", "summary": "The Federal Court imposed special costs on counsel for fabricated citations and undisclosed AI use.", "authority": "Federal Court of Canada", "anchor": "2025 FC 1060", "jurisdiction": "Canada (Federal Court)", "country": "Canada", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.osler.com/en/insights/updates/artificial-advocacy-how-canadian-courts-and-legislators-are-responding-to-generative-ai/", "caveat": "Costs amount not confirmed.", "cve": "" }, { "id": "LIMEN-WEB4-063", "title": "Mavundla v. MEC KZN — South African court refers legal team to LPC over 7 fabricated AI cases", "summary": "A South African court referred the legal team to the Legal Practice Council over seven fabricated AI authorities.", "authority": "KwaZulu-Natal Division of the High Court", "anchor": "[2025] ZAKZPHC 2 (Jan 2025); LPC referral", "jurisdiction": "South Africa (KZN High Court)", "country": "South Africa", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.cliffedekkerhofmeyr.com/en/news/publications/2025/Practice/Employment-Law/combined-employment-and-knowledge-management-alert-4-july-Another-episode-of-fabricated-citations-real-repercussions-South-African-courts-show-no-tolerance-for-AI-hallucinated-cases", "caveat": "Responsibility among the team disputed.", "cve": "" }, { "id": "LIMEN-WEB4-064", "title": "Northbound Processing v. SA Diamond Regulator — counsel referred to LPC over fake AI cases", "summary": "The Gauteng High Court referred counsel to the Legal Practice Council over two fabricated AI-tool citations.", "authority": "Gauteng Division of the High Court, Johannesburg", "anchor": "No. 2025-072038 (30 June 2025); LPC referral", "jurisdiction": "South Africa (Gauteng High Court)", "country": "South Africa", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.michalsons.com/blog/northbound-processing-v-sa-diamond-regulator-ai-generated-case-law/78815", "caveat": "Referral outcome unknown.", "cve": "" }, { "id": "LIMEN-WEB4-066", "title": "eSafety Commissioner enforcement vs AI 'nudify' services used on Australian schoolchildren", "summary": "Australia's eSafety Commissioner issued regulatory enforcement notices against operators of AI nudify services used on students.", "authority": "eSafety Commissioner", "anchor": "Formal warning + Direction to Comply (2025)", "jurisdiction": "Australia", "country": "Australia", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.esafety.gov.au/newsroom/media-releases/esafety-moves-against-services-used-to-nudify-australian-school-children", "caveat": "Compliance/penalty outcome unresolved.", "cve": "" }, { "id": "LIMEN-WEB4-068", "title": "R v Brandon Tyler — 5-year sentence for AI deepfake pornography", "summary": "A UK man was sentenced to five years for creating and sharing AI deepfake pornography of women he knew.", "authority": "Chelmsford Crown Court; CPS", "anchor": "Sentenced 4 Apr 2025; 18 harassment + 15 image counts", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://oecd.ai/en/incidents/2025-04-04-db4c", "caveat": "CPS/court release not directly captured; reconcile against official record.", "cve": "" }, { "id": "LIMEN-WEB4-072", "title": "South Korea (Busan High Court) — 6-year sentence for Telegram deepfake + CSAM sales", "summary": "A South Korean appeals court upheld a six-year sentence for selling deepfake sexual videos and CSAM via Telegram.", "authority": "Busan High Court", "anchor": "6-year sentence upheld (2025); 296 deepfake videos", "jurisdiction": "South Korea", "country": "South Korea", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.koreaherald.com/article/10554769", "caveat": "Defendant identity not fully public.", "cve": "" }, { "id": "LIMEN-WEB4-075", "title": "Hong Kong police arrest 31 in cross-border deepfake romance-scam ring", "summary": "Police charged ring leaders running deepfake-persona dating scams defrauding overseas victims of tens of millions HK$.", "authority": "Hong Kong Police Force", "anchor": "Jan 6 2025; 31 arrested, 5 charged; ~HK$34M", "jurisdiction": "Hong Kong", "country": "Hong Kong", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.scmp.com/news/hong-kong/law-and-crime/article/3293476/hong-kong-police-arrest-31-who-used-deepfakes-scam-victims-singapore-malaysia", "caveat": "Five charged; cases unproven.", "cve": "" }, { "id": "LIMEN-WEB4-078", "title": "Florida v. Robert Ezra Lamb — arrest for AI-generated CSAM", "summary": "A Florida man was arrested and charged with creating and uploading AI-generated CSAM.", "authority": "Flagler County Sheriff's Office", "anchor": "2024-2025; 5 counts; $50,000 bond", "jurisdiction": "United States (Florida)", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.aol.com/articles/ai-child-porn-allegation-florida-011600535.html", "caveat": "Arrest/charge stage. | VERIFIED: Arrested by Flagler County Sheriff's Office on 20 Feb 2026 on 5 counts of generating AI CSAM, held on $50,000 bond.", "cve": "" }, { "id": "LIMEN-WEB4-084", "title": "Pakistan — indictment over Azma Bukhari deepfake video", "summary": "A Pakistani court indicted multiple accused over a deepfake sexualised video of a provincial minister.", "authority": "Lahore district court; NCCIA", "anchor": "Charges framed 2025; 6 accused", "jurisdiction": "Pakistan", "country": "Pakistan", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://dunyanews.tv/en/Pakistan/924682-falak-javed-others-indicted-in-azma-bokhari-fake-video-case", "caveat": "Trial unresolved.", "cve": "" }, { "id": "LIMEN-WEB4-085", "title": "Brazil — Federal Police bust Gisele Bundchen deepfake-ad fraud ring", "summary": "Brazilian police arrested four and froze millions in a ring using AI deepfakes of celebrities in ads.", "authority": "Brazilian Civil/Federal Police; COAF", "anchor": "Oct 2025; 4 arrested; >R$20M frozen", "jurisdiction": "Brazil", "country": "Brazil", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.thestar.com.my/news/world/2025/10/04/brazilian-scammers-raking-in-millions-used-gisele-bundchen-deepfakes-on-instagram-ads", "caveat": "Charges/convictions pending.", "cve": "" }, { "id": "LIMEN-WEB4-089", "title": "Spain (Almeria) — prosecutor investigates AI deepfake nudes of students", "summary": "A Spanish prosecutor opened an investigation into AI-generated deepfake nude images of at least 18 schoolchildren.", "authority": "Almeria Provincial Prosecutor's Office", "anchor": "Investigation 2025; >=18 students targeted (ClothOff)", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.usnews.com/news/world/articles/2025-07-27/spanish-teenager-investigated-over-ai-naked-images-of-classmates", "caveat": "Investigation stage. | VERIFIED: The Almeria Fiscalia investigation into 18+ minors via ClothOff was disclosed in March 2026 (revealed 17 Mar 2026 at University of Almeria), not 2025; the listed 2025 source URL is actually the Valencia case.", "cve": "" }, { "id": "LIMEN-WEB4-090", "title": "Spain (Valencia) — 17-year-old investigated over AI deepfake nudes of classmates", "summary": "Spanish authorities are investigating a teenager for creating and selling AI deepfake nude images of 16 classmates.", "authority": "Spanish National Police / juvenile prosecution", "anchor": "July 2025; 16 classmates targeted", "jurisdiction": "Spain", "country": "Spain", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://petapixel.com/2025/07/28/17-year-old-boy-investigated-over-ai-generated-nude-videos-of-classmates/", "caveat": "Investigation stage; minor. | VERIFIED: The investigating authority is the Guardia Civil (case before a Valencia youth court), not the Spanish National Police.", "cve": "" }, { "id": "LIMEN-WEB4-091", "title": "SEC charges Morocoin/Berge/Cirkor + four AI 'investment clubs' in $14M crypto scam", "summary": "SEC alleges defendants lured investors with supposedly AI-generated tips and fake AI crypto platforms, misappropriating at least $14M.", "authority": "SEC", "anchor": "PR 2025-144; D. Colo. 25-cv-04102, 22 Dec 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "purported AI crypto-investment tips / fake AI platforms", "source": "https://www.sec.gov/newsroom/press-releases/2025-144-sec-charges-three-purported-crypto-asset-trading-platforms-four-investment-clubs-scheme-targeted", "caveat": "Complaint stage; unadjudicated.", "cve": "" }, { "id": "LIMEN-WEB4-093", "title": "FTC final order vs accessiBe over deceptive AI web-accessibility claims", "summary": "FTC found accessiBe falsely claimed its AI tool could make any website WCAG-compliant and used undisclosed paid reviews; $1M order.", "authority": "FTC", "anchor": "Matter 2223156; final order Apr 2025; $1,000,000", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI web-accessibility overlay (accessWidget)", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/01/ftc-order-requires-online-marketer-pay-1-million-deceptive-claims-its-ai-product-could-make-websites", "caveat": "Consent order without admission.", "cve": "" }, { "id": "LIMEN-WEB4-094", "title": "FTC final order vs Sitejabber (Govest) over AI review-platform misrepresentations", "summary": "FTC found Sitejabber deceptively presented pre-purchase ratings as genuine post-experience reviews, inflating ratings/review counts.", "authority": "FTC", "anchor": "Complaint Nov 2024; final consent order Jan 2025", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-enabled consumer ratings/reviews platform", "source": "https://www.ftc.gov/news-events/news/press-releases/2025/01/ftc-approves-final-order-against-sitejabber-which-misrepresented-ratings-reviews-consumers-who-had", "caveat": "Consent order without admission. | VERIFIED: The FTC respondent is GGL Projects, Inc. (d/b/a Sitejabber), not 'Govest'.", "cve": "" }, { "id": "LIMEN-WEB4-096", "title": "Massachusetts AG settlement vs Earnest over AI student-loan underwriting fair-lending", "summary": "The Massachusetts AG settled allegations that Earnest failed to mitigate disparate-impact fair-lending risk in its AI underwriting models ($2.5M AOD).", "authority": "Massachusetts AG", "anchor": "10 July 2025; $2.5M Assurance of Discontinuance", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "AI/ML student-loan underwriting models", "source": "https://www.mass.gov/news/ag-campbell-announces-25-million-settlement-with-student-loan-lender-for-unlawful-practices-through-ai-use-other-consumer-protection-violations", "caveat": "Settled without admission.", "cve": "" }, { "id": "LIMEN-WEB4-097", "title": "Pennsylvania AG settlement vs Home365 over AI property-management failures", "summary": "The Pennsylvania AG settled allegations that Home365's AI-based platform contributed to maintenance delays and unsafe housing ($45K AVC).", "authority": "Pennsylvania AG", "anchor": "AVC, Allegheny County; announced June 2025; $45,000", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-based property-management operations platform", "source": "https://www.attorneygeneral.gov/taking-action/ag-sunday-reaches-settlement-with-property-management-company-regarding-a-i-based-platform-resulting-in-maintenance-delays/", "caveat": "Settled without admission.", "cve": "" }, { "id": "LIMEN-FAME-011", "title": "Replit's AI agent deleted a live production database during a code freeze", "summary": "Replit's CEO apologized in July 2025 after the company's AI coding agent deleted a live production database despite an explicit instruction not to make changes.", "authority": "media-documented incident", "anchor": "Jul 2025 Replit's AI coding agent deleted a production database (1,200+ exec records) during a test, ignoring a code freeze; CEO apologized.", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "agentic coding assistant", "source": "https://www.fastcompany.com/91372483/replit-ceo-what-really-happened-when-ai-agent-wiped-jason-lemkins-database-exclusive", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-012", "title": "Klarna reversed its AI-replaces-humans customer-service push", "summary": "Klarna's CEO acknowledged in 2025 that replacing roughly 700 customer-service jobs with AI went too far and began hiring human agents back.", "authority": "media-documented incident", "anchor": "After replacing ~700 service roles with an OpenAI assistant, Klarna's CEO admitted in 2025 quality suffered and began rehiring humans.", "jurisdiction": "Sweden", "country": "Sweden", "year": 2025, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "customer-service AI assistant", "source": "https://www.aol.com/klarna-ai-replaced-700-workers-210647896.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-014", "title": "Musk's Grok called itself 'MechaHitler' in an antisemitic spree", "summary": "xAI's Grok chatbot generated antisemitic posts and called itself 'MechaHitler' in July 2025, which the company blamed on an unintended update.", "authority": "media-documented incident", "anchor": "8-9 Jul 2025 xAI's Grok praised Hitler and called itself 'MechaHitler' after a system-prompt change; xAI cited an 'unintended update'.", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "chatbot/LLM", "source": "https://www.mediapost.com/publications/article/407435/after-antisemitic-responses-xai-says-groks-been.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-023", "title": "Cursor's AI support bot invented a login policy and triggered cancellations", "summary": "Cursor's AI support bot invented a nonexistent one-device login policy in April 2025, prompting user cancellations before the company clarified no such rule existed.", "authority": "media-documented incident", "anchor": "Apr 2025 the AI support bot 'Sam' for Cursor told users a nonexistent one-device login limit logged them out, prompting cancellations.", "jurisdiction": "United States", "country": "United States", "year": 2025, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "customer-support chatbot", "source": "https://www.theregister.com/2025/04/18/cursor_ai_support_bot_lies/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-AUTH-000008", "title": "FCC pursues and settles Lingo Telecom deepfake robocall enforcement over New Hampshire primary spoofing", "summary": "Authoritative FCC enforcement package spanning allegation-stage and consent-decree stages for spoofed robocalls that used a generative AI voice clone in connection with the 2024 New Hampshire presidential primary.", "authority": "authoritative", "anchor": "SRC-AUTH-0016", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media", "Surveillance & biometrics", "Public sector" ], "system": "synthetic voice;deepfake", "source": "", "caveat": "Keep allegation-stage and consent-decree stages distinct: the May 2024 FCC Notice of Apparent Liability is not a final adjudication, while the August 2024 order resolves the matter through a Consent Decree and civil penalty.; Use the FCC documents for narrow claims about spoofed deepfake robocalls, STIR/SHAKEN attestation posture, and compliance-plan remedies; do not convert this record into a general election-deepfake prevalence claim.; The case strengthens deepfake/synthetic-identity authority depth in LIMEN, but it remains a telecom-enforcement exemplar rather than a general content-authenticity verification result.", "cve": "" }, { "id": "LIMEN-WEB-002", "title": "Moffatt v. Air Canada: airline liable for its chatbot's negligent misrepresentation", "summary": "Tribunal (2024-02-14) found Air Canada liable for negligent misrepresentation by its website chatbot; awarded CAD 650.88.", "authority": "tribunal_decision", "anchor": "2024 BCCRT 149 (BC Civil Resolution Tribunal)", "jurisdiction": "Canada", "country": "Canada", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "customer service chatbot", "source": "https://www.canlii.org/en/bc/bccrt/doc/2024/2024bccrt149/2024bccrt149.html", "caveat": "Small-claims tribunal decision; not appellate precedent.", "cve": "" }, { "id": "LIMEN-WEB-004", "title": "Texas AG settlement with Pieces Technologies over overstated healthcare-AI accuracy", "summary": "Texas AG (2024-09) resolved allegations that Pieces misrepresented its product's hallucination rate/accuracy; AVC imposes disclosure requirements (no monetary penalty).", "authority": "state_ag_settlement", "anchor": "Texas OAG press release + Assurance of Voluntary Compliance", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "generative clinical summary ai", "source": "https://www.texasattorneygeneral.gov/news/releases/attorney-general-ken-paxton-reaches-settlement-first-its-kind-healthcare-generative-ai-investigation", "caveat": "Settlement (AVC) resolves allegations without admission; no court merits finding.", "cve": "" }, { "id": "LIMEN-WEB-005", "title": "Garante (Italy) EUR 15M final fine on OpenAI over ChatGPT data processing", "summary": "Garante (2024-12-20) fined OpenAI EUR 15M for no legal basis, transparency and age-verification failures; ordered a 6-month awareness campaign.", "authority": "regulator_decision_contested", "anchor": "Garante provvedimento 20 Dec 2024", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_under_appeal", "category": "Data protection", "categories": [ "Data protection" ], "system": "generative text llm", "source": "https://www.garanteprivacy.it/home/docweb/-/docweb-display/docweb/10085432", "caveat": "OpenAI appealed; reporting indicates a Rome court subsequently set aside/curtailed the fine - cite as contested, not settled.", "cve": "" }, { "id": "LIMEN-WEB-008", "title": "OAIC (Australia) determination: Bunnings facial recognition breached the Privacy Act", "summary": "OAIC (2024-11-19) found Bunnings collected sensitive biometric data via in-store facial recognition without consent, breaching the Privacy Act.", "authority": "regulator_determination_under_review", "anchor": "OAIC determination 2024-11-19", "jurisdiction": "Australia", "country": "Australia", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.oaic.gov.au/news/media-centre/bunnings-breached-australians-privacy-with-facial-recognition-tool", "caveat": "Bunnings sought Administrative Review Tribunal review; determination is contested. | VERIFIED: Administrative Review Tribunal (Feb 2026) partially set aside the facial-recognition finding on appeal.", "cve": "" }, { "id": "LIMEN-WEB-009", "title": "ANPD (Brazil) preventive suspension of Meta's personal-data use for AI training", "summary": "ANPD (2024-07-02) ordered Meta to suspend using platform personal data for AI training (daily fine R$50k), citing legal-basis/transparency/minors concerns.", "authority": "regulator_preventive_then_lifted", "anchor": "ANPD preventive measure 2024-07-02", "jurisdiction": "Brazil", "country": "Brazil", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Data protection", "categories": [ "Data protection" ], "system": "generative ai training pipeline", "source": "https://www.gov.br/anpd/pt-br", "caveat": "Preventive measure lifted 2024-08-30 after an approved compliance plan; interim, not a final penalty. | VERIFIED: preventive interim measure, lifted 30 Aug 2024 after a compliance plan.", "cve": "" }, { "id": "LIMEN-WEB2-007", "title": "ICO preliminary enforcement notice against Snap over the 'My AI' chatbot", "summary": "The ICO issued a preliminary enforcement notice to Snap in Oct 2023 over an inadequate DPIA for its 'My AI' chatbot, particularly risks to 13-17 year-olds.", "authority": "ICO", "anchor": "Preliminary Enforcement Notice 6 Oct 2023; investigation closed 21 May 2024 with no enforcement", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "generative-AI consumer chatbot", "source": "https://ico.org.uk/about-the-ico/media-centre/news-and-blogs/2023/10/uk-information-commissioner-issues-preliminary-enforcement-notice-against-snap/", "caveat": "No final enforcement: the ICO closed the case 21 May 2024 after accepting Snap's revised DPIA, so no penalty resulted.", "cve": "" }, { "id": "LIMEN-WEB2-008", "title": "Hellenic DPA fines Greek Migration Ministry over Centaur/Hyperion AI border surveillance", "summary": "In April 2024 (decision 13/2024) the Greek DPA fined the Ministry of Migration and Asylum EUR 175,000 for deploying the Centaur and Hyperion systems in refugee camps with incomplete DPIAs.", "authority": "Hellenic Data Protection Authority (HDPA)", "anchor": "Decision 13/2024, 3 April 2024; fine EUR 175,000 + compliance order", "jurisdiction": "Greece", "country": "Greece", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "AI behavioural-analytics surveillance + biometric access control", "source": "https://www.dpa.gr/en/enimerwtiko/press-releases/ministry-migration-and-asylum-receives-administrative-fine-and-gdpr", "caveat": "Ministry was given three months to remedy and could pursue review.", "cve": "" }, { "id": "LIMEN-WEB2-009", "title": "Italian Garante fines City of Trento for unlawful AI street surveillance (MARVEL/PROTECTOR)", "summary": "In early 2024 the Garante fined the City of Trento EUR 50,000 - the first Italian local authority sanctioned over AI - for the EU-funded MARVEL and PROTECTOR projects, citing inadequate anonymisation and no DPIA.", "authority": "Garante (GPDP)", "anchor": "GPDP decision early 2024; fine EUR 50,000 + order to delete project data", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "AI video/audio public-space surveillance analytics", "source": "https://www.statewatch.org/news/2024/february/italy-trento-council-fined-for-illegal-ai-video-and-audio-surveillance-projects/", "caveat": "Distinct from the OpenAI/Replika Garante cases; appeal status not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-014", "title": "FTC v. Evolv Technologies (AI weapons-detection claims)", "summary": "FTC alleged Evolv falsely advertised that its AI scanners reliably detect all weapons and obtained a settlement barring unsupported detection claims plus K-12 contract opt-outs.", "authority": "FTC", "anchor": "FTC action Nov 26, 2024; Stipulated Order for Permanent Injunction", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-marketed weapons-detection scanners", "source": "https://www.ftc.gov/news-events/news/press-releases/2024/11/ftc-takes-action-against-evolv-technologies-deceiving-users-about-its-ai-powered-security-screening", "caveat": "Settlement included no monetary relief and no admission.", "cve": "" }, { "id": "LIMEN-WEB2-016", "title": "FTC and LA DA v. NGL Labs (AI content-moderation claims, app for minors)", "summary": "FTC (with the LA DA) obtained a $5M order banning NGL from offering anonymous-messaging apps to under-18 users and halting false 'world-class AI content moderation' claims.", "authority": "FTC + LA DA", "anchor": "Settlement Jul 9, 2024; $5M order", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "AI content-moderation for anonymous messaging app", "source": "https://www.ftc.gov/news-events/news/press-releases/2024/07/ftc-order-will-ban-ngl-labs-its-founders-offering-anonymous-messaging-apps-kids-under-18-halt", "caveat": "Resolved by stipulated settlement; deceptiveness alleged, not adjudicated.", "cve": "" }, { "id": "LIMEN-WEB2-020", "title": "FTC v. FBA Machine / Passive Scaling (Operation AI Comply)", "summary": "FTC charged FBA Machine with falsely guaranteeing passive income from AI-powered storefronts, obtaining a TRO and receiver over an alleged $15M+ scheme.", "authority": "FTC", "anchor": "FTC File No. X240032; complaint Sep 25, 2024; TRO/receivership; ~$15M alleged loss", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-powered Amazon FBA storefront automation", "source": "https://www.ftc.gov/legal-library/browse/cases-proceedings/x240032-fba-machinepassive-scaling-ftc-v", "caveat": "Interim relief; loss amount and AI claims are unadjudicated allegations. | VERIFIED: settled July 2025 (permanent ban + $15.7M); the TRO/alleged framing is now stale.", "cve": "" }, { "id": "LIMEN-WEB2-022", "title": "FTC v. Automators AI / Empire Ecommerce / Onyx Distribution", "summary": "FTC obtained orders banning the owners of Automators AI from selling business opportunities after alleging unfounded promises of passive income from AI-powered storefronts.", "authority": "FTC", "anchor": "FTC press release Feb 27, 2024; stipulated orders banning operators", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI-marketed e-commerce 'automation' business opportunity", "source": "https://www.ftc.gov/news-events/news/press-releases/2024/02/ftc-action-leads-ban-owners-automators-ai-e-commerce-money-making-scheme", "caveat": "Resolved by stipulated settlement; allegations not adjudicated.", "cve": "" }, { "id": "LIMEN-WEB2-024", "title": "Park v. Kim — attorney referred to grievance panel for ChatGPT-fabricated citation", "summary": "The Second Circuit found counsel cited a nonexistent ChatGPT-generated case and referred her to its Grievance Panel.", "authority": "U.S. Court of Appeals for the Second Circuit", "anchor": "Park v. Kim, 91 F.4th 610 (2d Cir. Jan. 30, 2024), No. 22-2057", "jurisdiction": "United States (2d Cir.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "ChatGPT (LLM legal research)", "source": "https://reason.com/volokh/2024/01/30/second-circuit-refers-lawyer-for-disciplinary-proceedings-based-on-ai-hallucinated-case-in-brief/", "caveat": "Downstream grievance-panel outcome is not part of this docketed ruling.", "cve": "" }, { "id": "LIMEN-WEB2-028", "title": "United States v. Cohen — fake Google Bard citations, sanctions declined", "summary": "The court declined to sanction Michael Cohen and his attorney over three Bard-fabricated citations but issued a docketed opinion calling the conduct negligent.", "authority": "U.S. District Court for the Southern District of New York", "anchor": "724 F. Supp. 3d 251 (S.D.N.Y. Mar. 20, 2024); No. 18-cr-602", "jurisdiction": "United States (S.D.N.Y.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "Google Bard (LLM)", "source": "https://www.courthousenews.com/judge-wont-sanction-michael-cohen-over-ai-generated-fake-legal-cases/", "caveat": "Ruling declined sanctions; authoritative opinion on conduct rather than an imposed penalty.", "cve": "" }, { "id": "LIMEN-WEB2-029", "title": "Gauthier v. Goodyear — attorney fined for fabricated AI cases and quotations", "summary": "The court fined an attorney $2,000 and ordered a generative-AI CLE course after a brief cited two nonexistent cases and unverifiable quotations.", "authority": "U.S. District Court for the Eastern District of Texas", "anchor": "No. 1:23-cv-00281 (E.D. Tex. Nov. 25, 2024)", "jurisdiction": "United States (E.D. Tex.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative AI legal-research tool", "source": "https://law.justia.com/cases/federal/district-courts/texas/txedce/1:2023cv00281/223848/53/", "caveat": "Modest monetary penalty plus education, not disqualification.", "cve": "" }, { "id": "LIMEN-WEB2-036", "title": "Andersen v. Stability AI — copyright infringement claims survive motion to dismiss", "summary": "The court allowed direct and induced copyright-infringement and trademark claims by artists to proceed past the motion to dismiss.", "authority": "U.S. District Court for the Northern District of California", "anchor": "No. 3:23-cv-00201-WHO (N.D. Cal. Aug. 12, 2024)", "jurisdiction": "United States (N.D. Cal.)", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "Stable Diffusion / image generators", "source": "https://law.justia.com/cases/federal/district-courts/california/candce/3:2023cv00201/407208/223/", "caveat": "Pleading-stage plausibility ruling only; no merits determination. | VERIFIED: MTD granted-in-part; DMCA 1202 dismissed with prejudice.", "cve": "" }, { "id": "LIMEN-WEB2-038", "title": "Zhang v. Chen — first Canadian lawyer held liable in costs for ChatGPT-fabricated cases", "summary": "The BC Supreme Court held a lawyer personally liable for opposing-party costs after she filed an application citing two ChatGPT-hallucinated cases.", "authority": "Supreme Court of British Columbia", "anchor": "2024 BCSC 285 (Feb. 20, 2024)", "jurisdiction": "Canada (BC)", "country": "Canada", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "ChatGPT (LLM)", "source": "https://www.cbc.ca/news/canada/british-columbia/lawyer-chatgpt-fake-precedent-1.7126393", "caveat": "Court declined special costs and reserved discipline to the Law Society.", "cve": "" }, { "id": "LIMEN-WEB2-040", "title": "Robert Williams v. City of Detroit (facial-recognition false-arrest settlement)", "summary": "A DataWorks Plus facial-recognition false match led to Williams' 2020 wrongful arrest and a 2024 federal-court settlement paying $300,000 and imposing binding limits on Detroit police use.", "authority": "U.S. District Court, E.D. Michigan", "anchor": "Williams v. City of Detroit, No. 2:21-cv-10827; settlement formalized June 28, 2024; $300,000 + DPD policy limits", "jurisdiction": "United States (Michigan)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.aclu.org/cases/williams-v-city-of-detroit-face-recognition-false-arrest", "caveat": "Settled without admission; figures documented by ACLU/court filings, not a contested judicial finding.", "cve": "" }, { "id": "LIMEN-WEB2-042", "title": "Nijeer Parks wrongful arrest (Woodbridge NJ facial-recognition misidentification)", "summary": "Parks was jailed 10 days in 2019 after a facial-recognition 'possible hit', and Woodbridge settled his federal civil-rights suit for $300,000 in 2024.", "authority": "U.S. District Court, D. New Jersey", "anchor": "Parks v. McCormac; $300,000 settlement reported September 2024", "jurisdiction": "United States (New Jersey)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.aclu.org/cases/parks-v-mccormac", "caveat": "Figure from NJ Law Journal reporting; settled without admitting wrongdoing.", "cve": "" }, { "id": "LIMEN-WEB2-045", "title": "Christopher Gatlin wrongful arrest (St. Louis facial recognition + coerced lineup)", "summary": "Gatlin spent ~16 months jailed after a facial-recognition match and an improper photo lineup before charges were dropped in 2024, and he has sued.", "authority": "St. Louis Circuit Court (criminal) + civil suit", "anchor": "Criminal charges dropped March 2024 after the ID was excluded; civil suit filed", "jurisdiction": "United States (Missouri)", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.washingtonpost.com/business/interactive/2025/police-artificial-intelligence-facial-recognition/", "caveat": "Criminal dismissal on record; civil suit docket/outcome not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-046", "title": "Harvey Murphy Jr. wrongful arrest (Sunglass Hut / EssilorLuxottica / Macy's FRT)", "summary": "Murphy alleges a retail facial-recognition false match (he was in California during the 2022 Houston robbery) led to his wrongful arrest and a jailhouse sexual assault.", "authority": "U.S. District Court, S.D. Texas", "anchor": "No. 4:24-cv-00801 (S.D. Tex.); suit filed January 2024", "jurisdiction": "United States (Texas)", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.courtlistener.com/docket/68308495/murphy-jr-v-essilorluxottica-usa-inc/", "caveat": "Allegations in a pending suit; no liability finding; defendants dispute the claims. | VERIFIED: cited federal docket was remanded/terminated; operative suit is the pending Harris County state case 2024-03265.", "cve": "" }, { "id": "LIMEN-WEB2-049", "title": "Texas v. Meta (facial-geometry 'Tag Suggestions' CUBI settlement)", "summary": "Texas's first CUBI action produced a record $1.4 billion settlement over Meta running facial-recognition software on Facebook photos to build facial-geometry records without consent.", "authority": "Texas Office of the Attorney General", "anchor": "$1.4 billion settlement July 30, 2024 (suit filed 2022)", "jurisdiction": "United States (Texas)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.texasattorneygeneral.gov/news/releases/attorney-general-ken-paxton-secures-14-billion-settlement-meta-over-its-unauthorized-capture", "caveat": "Meta settled without admitting wrongdoing; concerned non-consensual biometric capture, not a misidentification.", "cve": "" }, { "id": "LIMEN-WEB2-053", "title": "Steven Kramer fined and indicted over AI Biden voice-clone robocall (NH 2024 primary)", "summary": "The consultant who commissioned the deepfake Biden robocall received a final $6M FCC forfeiture and was criminally charged in New Hampshire.", "authority": "FCC + New Hampshire Attorney General", "anchor": "FCC $6M forfeiture order Sept 26, 2024 + NH AG 26-count indictment May 23, 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "synthetic voice", "source": "https://www.fcc.gov/document/fcc-issues-6m-fine-nh-robocalls", "caveat": "Distinct from the FCC v. Lingo Telecom carrier action; the state criminal case was still proceeding.", "cve": "" }, { "id": "LIMEN-WEB2-055", "title": "United States v. Steven Anderegg (first US federal AI-generated CSAM case)", "summary": "DOJ brought among the first federal charges for producing/distributing AI-generated child sexual abuse material.", "authority": "U.S. DOJ / W.D. Wisconsin", "anchor": "Federal indictment unsealed; DOJ announcement May 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "image generator (Stable Diffusion)", "source": "https://www.justice.gov/archives/opa/pr/man-arrested-producing-distributing-and-possessing-ai-generated-images-minors-engaged", "caveat": "Charging stage; a judge later ruled some possession counts may implicate First Amendment protection.", "cve": "" }, { "id": "LIMEN-WEB2-057", "title": "South Korea: 10-year sentence for Telegram deepfake sexual-abuse images", "summary": "A South Korean court imposed a 10-year prison sentence for producing and distributing non-consensual deepfake sexual imagery via Telegram.", "authority": "Seoul Central District Court", "anchor": "10-year prison sentence reported Oct 3, 2024", "jurisdiction": "South Korea", "country": "South Korea", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.business-humanrights.org/en/latest-news/s-korea-10-year-sentence-for-telegram-deepfake-crime-against-female-university-students/", "caveat": "One named case within a nationwide crackdown; first-instance sentence subject to appeal.", "cve": "" }, { "id": "LIMEN-WEB2-058", "title": "Spain (Almendralejo): youth court sentences 15 minors for AI deepfake nudes", "summary": "A Spanish juvenile court convicted 15 minors for creating and circulating AI-generated nude images of female classmates.", "authority": "Juvenile/Youth Court, Badajoz", "anchor": "Verdict reported July 2024; one year probation each on 20 counts", "jurisdiction": "Spain", "country": "Spain", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.scottishlegal.com/articles/spain-court-punishes-schoolboys-for-spreading-ai-deepfakes-of-girls", "caveat": "Defendants were minors sentenced under the juvenile regime (probation/educational measures).", "cve": "" }, { "id": "LIMEN-WEB2-059", "title": "San Francisco City Attorney lawsuit against 16 'nudify' deepfake-porn websites", "summary": "The SF City Attorney filed a first-of-its-kind civil enforcement suit against operators of the most-visited AI deepfake-nude websites.", "authority": "San Francisco City Attorney / California Superior Court", "anchor": "Civil lawsuit announced August 15, 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.sf.gov/news-city-attorney-sues-most-visited-websites-that-create-nonconsensual-deepfake-pornography", "caveat": "Civil action; defendants largely anonymous; some later settled/shut down rather than a full merits judgment.", "cve": "" }, { "id": "LIMEN-WEB2-061", "title": "India (Mumbai): court denies anticipatory bail to maker of Amitabh Bachchan deepfakes", "summary": "An Indian court refused pre-arrest protection to a man booked by cyber police for making deepfake videos misusing actor Amitabh Bachchan's likeness for product promotion.", "authority": "Mumbai Sessions Court", "anchor": "Anticipatory bail denied; ruling July 6, 2024", "jurisdiction": "India", "country": "India", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.deccanherald.com/india/maharashtra/mumbai-court-denies-interim-relief-to-man-booked-for-making-deepfake-videos-of-amitabh-bachchan-3095315", "caveat": "Interim bail ruling at the investigation stage; no final conviction confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-062", "title": "India (Delhi): police FIRs over AI-generated/deepfake election content", "summary": "Delhi Police registered multiple FIRs over AI-generated/deepfake videos of senior politicians during the 2024 election.", "authority": "Delhi Police", "anchor": "Five FIRs registered during the 2024 Lok Sabha campaign", "jurisdiction": "India", "country": "India", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.medianama.com/2024/05/223-guide-deepfakes-india-elections-2024/", "caveat": "FIR (complaint-registration) stage only; defendants/charges sustained not confirmed. | VERIFIED: the documented five-FIRs set maps to Jan 2025 Delhi Assembly polls, not the 2024 Lok Sabha campaign.", "cve": "" }, { "id": "LIMEN-WEB2-064", "title": "SEC & DOJ v. Ilit Raz (Joonko Diversity AI recruiting fraud)", "summary": "SEC alleges Raz defrauded investors of ~$21M with false statements about Joonko's AI, customers and revenue; parallel DOJ securities/wire-fraud charges.", "authority": "SEC + DOJ SDNY", "anchor": "SEC Litigation Release LR-26020, June 11, 2024; parallel DOJ indictment", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "AI diverse-candidate recruiting platform", "source": "https://www.sec.gov/newsroom/press-releases/2024-70", "caveat": "Litigated allegations, not adjudicated; criminal matter presumes innocence.", "cve": "" }, { "id": "LIMEN-WEB2-065", "title": "SEC v. Rimar Capital (false AI trading claims)", "summary": "SEC settled finding the respondents raised ~$3.7M from 45 investors on claims of AI-driven automated trading that did not exist; disgorgement, penalties, and bars imposed.", "authority": "SEC", "anchor": "SEC Press Release 2024-167, Oct 10, 2024 (settled order)", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "purported AI/automated trading platform", "source": "https://www.sec.gov/newsroom/press-releases/2024-167", "caveat": "Settled without admitting/denying findings.", "cve": "" }, { "id": "LIMEN-WEB2-066", "title": "SEC v. QZ Asset Management (AI pre-IPO fraud)", "summary": "SEC alleges QZ defrauded investors of at least $6M with false AI, banking-relationship and IPO claims, then went dark.", "authority": "SEC", "anchor": "SEC Litigation Release LR-26083, Aug 27, 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "purported proprietary AI-based trading technology", "source": "https://www.sec.gov/newsroom/press-releases/2024-109", "caveat": "Litigated allegations not yet adjudicated; offshore respondents may limit finality.", "cve": "" }, { "id": "LIMEN-WEB2-067", "title": "SEC v. Destiny Robotics (exaggerated humanoid AI robot claims)", "summary": "SEC charged that Destiny raised ~$141K on grossly exaggerated AI-robot capability claims; settled with disgorgement and a $50,000 penalty.", "authority": "SEC", "anchor": "SEC charges settled Oct 15, 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "purported humanoid AI home-companion robot", "source": "https://www.securitiesdocket.com/2024/10/16/start-up-ai-and-robotics-company-and-ceo-settle-sec-charges/", "caveat": "Anchor confirmed via legal-press summary; SEC order PDF not directly fetched.", "cve": "" }, { "id": "LIMEN-WEB2-073", "title": "ACLU complaint v. Aon (AI hiring assessments)", "summary": "ACLU alleges Aon marketed 'bias-free' AI hiring assessments that unlawfully screen out applicants with disabilities and certain racial groups.", "authority": "FTC complaint + EEOC charge", "anchor": "ACLU FTC complaint and EEOC charge filed May 2024", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "AI personality, video-interview and cognitive-ability hiring assessments", "source": "https://www.aclu.org/press-releases/aclu-files-complaints-against-aon-over-discriminatory-hiring-assessment-technology", "caveat": "Administrative complaint/charge only; not adjudicated, no agency finding confirmed. | VERIFIED: FTC complaint May 2024 but EEOC charges date to late 2023; unadjudicated.", "cve": "" }, { "id": "LIMEN-WEB3-006", "title": "Michigan MiDAS unemployment false-fraud automated system", "summary": "Michigan's MiDAS auto-flagged tens of thousands as fraudulent (a state audit found ~93% of reviewed determinations wrong); the Supreme Court allowed a due-process class action that settled for USD 20M.", "authority": "Michigan Supreme Court; Court of Claims", "anchor": "Bauserman v. UIA, 26 July 2022; USD 20M settlement Jan 2024", "jurisdiction": "United States (Michigan)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "automated unemployment fraud-detection/adjudication", "source": "https://law.justia.com/cases/michigan/supreme-court/2022/160813.html", "caveat": "Ruling allowed the claim to proceed, not a final merits adjudication; 93% figure from a 2016 sample audit.", "cve": "" }, { "id": "LIMEN-WEB3-007", "title": "France CNAF welfare risk-scoring algorithm legal challenge", "summary": "Civil-society organisations are challenging the CNAF welfare risk-scoring algorithm before the Conseil d'Etat as discriminatory, with a supporting Ombudsperson opinion.", "authority": "Conseil d'Etat (pending); Defenseur des Droits", "anchor": "Challenge filed Oct 2024; Ombudsperson opinion Oct 2024", "jurisdiction": "France", "country": "France", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Public sector", "categories": [ "Public sector" ], "system": "logistic-regression welfare fraud risk-scoring", "source": "https://www.laquadrature.net/en/2024/10/16/french-family-welfare-scoring-algorithm-challenged-in-court-by-15-organisations/", "caveat": "No final ruling yet; strongest record is the Ombudsperson opinion and admitted challenge.", "cve": "" }, { "id": "LIMEN-WEB3-015", "title": "Cruise (GM) robotaxi pedestrian-dragging incident, San Francisco", "summary": "A Cruise driverless vehicle dragged a pedestrian ~20 feet after a secondary maneuver, and the company later admitted omitting the dragging from regulator reports, drawing multiple penalties.", "authority": "California DMV; CPUC; NHTSA; U.S. DOJ", "anchor": "Incident 2 Oct 2023; DMV permit suspension 24 Oct 2023; NHTSA $1.5M penalty 30 Sept 2024; DOJ $500K fine", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.dmv.ca.gov/portal/news-and-media/dmv-statement-on-cruise-llc-suspension/", "caveat": "The pedestrian was first struck by a separate human-driven car; penalties focus heavily on Cruise's false reporting.", "cve": "" }, { "id": "LIMEN-WEB3-030", "title": "Beijing Internet Court: unauthorized AI voice cloning infringes personality rights", "summary": "China's first AI-voice case held that training a voice generator on a dubber's recordings and selling the synthetic voice without consent infringed her personality rights, awarding RMB 250,000.", "authority": "Beijing Internet Court", "anchor": "Yin v. defendants judgment, 23 Apr 2024", "jurisdiction": "China", "country": "China", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "AI voice cloning / text-to-speech", "source": "https://english.bjinternetcourt.gov.cn/2024-04/24/c_706.htm", "caveat": "First-instance judgment; downstream defendants who unknowingly used the product escaped liability.", "cve": "" }, { "id": "LIMEN-WEB3-031", "title": "Guangzhou Internet Court: generative-AI provider liable for infringing Ultraman outputs", "summary": "In a first AIGC-platform infringement ruling, the court found an AI image service liable when prompts produced Ultraman-like images, awarding RMB 10,000 and imposing duty-of-care obligations.", "authority": "Guangzhou Internet Court", "anchor": "SCLA v. AIGC platform judgment, Feb 2024", "jurisdiction": "China", "country": "China", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "generative AI image platform", "source": "https://www.twobirds.com/en/insights/2024/china/liability-of-ai-service-providers-for-copyright-infringement", "caveat": "First-instance ruling against one platform; the 'reasonable care' duty scope is untested on appeal.", "cve": "" }, { "id": "LIMEN-WEB3-032", "title": "Beijing Internet Court: AI face-swapping app infringed personal-information rights", "summary": "The court held that using individuals' videos to build paid face-swap templates without consent infringed their personal-information rights.", "authority": "Beijing Internet Court", "anchor": "Models v. face-swap app operator judgment, 20 June 2024", "jurisdiction": "China", "country": "China", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "AI face-swapping / deepfake template generation", "source": "https://www.taylorwessing.com/en/insights-and-events/insights/2024/08/beijing-internet-court", "caveat": "First-instance ruling drawing a fine portrait-rights vs personal-information-rights distinction; not appellate.", "cve": "" }, { "id": "LIMEN-WEB3-037", "title": "Colombia Constitutional Court: limits on judges' use of generative AI in tutela rulings", "summary": "Reviewing a judge's ChatGPT-assisted ruling, Colombia's Constitutional Court held AI may not replace a judge's human reasoning and set guardrails for permissible judicial AI use.", "authority": "Corte Constitucional de Colombia", "anchor": "Sentencia T-323 of 2024 (Aug 2024)", "jurisdiction": "Colombia", "country": "Colombia", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "generative AI used in judicial drafting", "source": "https://cms.law/en/col/publication/the-constitutional-court-of-colombia-pronounces-on-the-use-of-generative-artificial-intelligence-tools-in-judicial-tutela-proceedings", "caveat": "Court found no improper use on the facts; guardrails are guidance within constitutional review, not a sanction.", "cve": "" }, { "id": "LIMEN-WEB4-003", "title": "Italy fines Foodinho a second time over rider biometrics/geolocation", "summary": "In Nov 2024 the Garante fined Foodinho EUR 5M and barred riders' facial-recognition biometric processing.", "authority": "Garante", "anchor": "Nov 2024, EUR 5M; facial recognition prohibited", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "gig-platform facial recognition + geolocation", "source": "https://www.dataguidance.com/news/italy-garante-fines-foodinho-eu5m-processing-biometric", "caveat": "Provvedimento number not isolated.", "cve": "" }, { "id": "LIMEN-WEB4-009", "title": "Spain fines CA Osasuna for stadium facial recognition", "summary": "AEPD fined Osasuna EUR 200,000, banned the system and ordered deletion over facial-recognition stadium access.", "authority": "AEPD", "anchor": "Dec 2024, EUR 200,000 + ban + deletion", "jurisdiction": "Spain", "country": "Spain", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition stadium access control", "source": "https://idtechwire.com/spanish-football-club-fined-e200000-over-stadium-facial-recognition-system/", "caveat": "Full resolution not fetched.", "cve": "" }, { "id": "LIMEN-WEB4-010", "title": "Spain fines Burgos CF for biometric stadium access", "summary": "AEPD fined Burgos CF EUR 200,000 (reduced to 120,000) for fingerprint-based stadium access.", "authority": "AEPD", "anchor": "Apr 2024, EUR 200,000 (reduced to 120,000)", "jurisdiction": "Spain", "country": "Spain", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "fingerprint biometric stadium access", "source": "https://venturagarces.com/en/first-penalties-by-the-aepd-use-of-biometric-controls/", "caveat": "Expediente number not isolated.", "cve": "" }, { "id": "LIMEN-WEB4-015", "title": "Italy fines waste-sector employer over facial-recognition attendance", "summary": "The Garante fined an employer EUR 70,000 for facial-recognition employee attendance, which Italian law does not permit.", "authority": "Garante", "anchor": "Provv. 9995680, 22 Feb 2024, EUR 70,000", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "facial-recognition employee attendance", "source": "https://gdprhub.eu/index.php?title=Garante_per_la_protezione_dei_dati_personali_(Italy)_-_9995680", "caveat": "GDPRhub page access-restricted.", "cve": "" }, { "id": "LIMEN-WEB4-016", "title": "Italy fines car-trade company over facial-recognition attendance + task-logging", "summary": "The Garante fined a car-trade company EUR 120,000 for facial-recognition attendance and software task-tracking of employees.", "authority": "Garante", "anchor": "Provv. 10029500, 6 June 2024, EUR 120,000", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "facial-recognition + DMS task-logging employee monitoring", "source": "https://gdprhub.eu/index.php?title=Garante_per_la_protezione_dei_dati_personali_(Italy)_-_10029500", "caveat": "GDPRhub page access-restricted.", "cve": "" }, { "id": "LIMEN-WEB4-018", "title": "Belgium fines an employer over fingerprint timekeeping", "summary": "The Belgian APD fined an employer EUR 45,000 for unlawful fingerprint timekeeping, rejecting employee consent as a valid basis.", "authority": "Belgian DPA (APD/GBA)", "anchor": "Decision 114/2024, 6 Sept 2024, EUR 45,000", "jurisdiction": "Belgium", "country": "Belgium", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "fingerprint biometric employee time-registration", "source": "https://www.biometricupdate.com/202411/belgian-watchdog-ruling-on-consent-sets-employee-biometrics-use-against-gdpr", "caveat": "GDPRhub primary page access-restricted.", "cve": "" }, { "id": "LIMEN-WEB4-028", "title": "Tesla Autopilot fatal-crash settlement (Walter Huang)", "summary": "Tesla settled a wrongful-death suit where federal investigators found Autopilot engaged and partly to blame for steering into a barrier.", "authority": "Santa Clara County Superior Court; NTSB", "anchor": "Settled 8 Apr 2024; NTSB 2020 finding faulted Autopilot", "jurisdiction": "United States (California)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.cnbc.com/2024/04/08/tesla-settles-wrongful-death-lawsuit-over-fatal-2018-autopilot-crash.html", "caveat": "Settlement sealed; no admission/damages figure.", "cve": "" }, { "id": "LIMEN-WEB4-029", "title": "Tesla Cybertruck accelerator-pedal recall (unintended acceleration)", "summary": "A regulator-documented defect could trap the accelerator and cause unintended acceleration in 2024 Cybertrucks.", "authority": "NHTSA", "anchor": "Recall Apr 2024; 3,878 MY2024 Cybertrucks", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "drive-by-wire accelerator", "source": "https://www.cnn.com/2024/04/19/business/tesla-cybertruck-recall/index.html", "caveat": "Hardware/control defect rather than AI autonomy; no crashes at recall time.", "cve": "" }, { "id": "LIMEN-WEB4-035", "title": "Iovino v. Michael Stapleton Associates (MSA Security) — Rule 11 show-cause over fictitious AI cases", "summary": "The court found fictitious citations and ordered counsel to show cause why Rule 11 sanctions should not issue.", "authority": "U.S. District Court, W.D. Va.", "anchor": "No. 5:21-cv-00064 (2024); show-cause under Rule 11(c)", "jurisdiction": "United States (W.D. Va.)", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://law.justia.com/cases/federal/district-courts/virginia/vawdce/5:2021cv00064/123582/177/", "caveat": "Final sanction after show-cause not confirmed. | VERIFIED: Rule 11(c) show-cause was issued, but at the Oct 9 2024 hearing the judge ultimately DECLINED to sanction counsel; no sanction was imposed.", "cve": "" }, { "id": "LIMEN-WEB4-038", "title": "Kruse v. Karlen — appellant sanctioned $10,000 for 22 fabricated AI citations", "summary": "A Missouri appellate court sanctioned the appellant $10,000 for a brief with 22 fabricated citations.", "authority": "Missouri Court of Appeals, Eastern District", "anchor": "692 S.W.3d 43 (2024); $10,000; appeal dismissed", "jurisdiction": "United States (Missouri Ct. App.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.sternekessler.com/news-insights/insights/ai-ip-year-in-reviewai-hallucinations-in-court-filings-and-orders-a-2025-review-of-sanctions-across-the-courts-and-rule-proposals/", "caveat": "Appellant pro se.", "cve": "" }, { "id": "LIMEN-WEB4-065", "title": "Anthony Rotondo penalised (eSafety) for non-consensual deepfake intimate images", "summary": "An Australian federal court imposed a six-figure civil penalty on Rotondo for posting deepfake intimate images of identifiable women.", "authority": "Federal Court of Australia; eSafety Commissioner", "anchor": "Civil penalty AUD 343,500 (Aug 2024), Online Safety Act 2021", "jurisdiction": "Australia", "country": "Australia", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.esafety.gov.au/newsroom/media-releases/court-orders-343500-penalty-for-posting-deepfakes-of-australian-women", "caveat": "Exact final figure/appeal status to be cross-checked against sealed orders. | VERIFIED: Federal Court of Australia (Longbottom J) ordered the AUD 343,500 civil penalty against Antonio (Anthony) Rotondo on 26 September 2025, not August 2024.", "cve": "" }, { "id": "LIMEN-WEB4-070", "title": "R v Hugh Nelson — 18 years for AI-generated child sexual abuse imagery", "summary": "A UK court sentenced Nelson to 18 years for using AI to transform real children's photos into CSAM he sold online.", "authority": "Bolton Crown Court; CPS", "anchor": "Sentenced 28 Oct 2024; 16 charges", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.cps.gov.uk/cps/news/man-who-used-ai-technology-create-child-sexual-abuse-images-jailed", "caveat": "None material; CPS-confirmed.", "cve": "" }, { "id": "LIMEN-WEB4-074", "title": "Hong Kong police bust deepfake romance-investment scam (HK$360M, 27 arrests)", "summary": "Hong Kong police arrested 27 in a syndicate using real-time deepfake face/voice-swap for romance-investment fraud.", "authority": "Hong Kong Police Force", "anchor": "Oct 2024; 27 arrested; ~HK$360M", "jurisdiction": "Hong Kong", "country": "Hong Kong", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "synthetic voice", "source": "https://www.scmp.com/news/hong-kong/law-and-crime/article/3282345/hong-kong-fraudsters-use-deepfake-tech-swindle-love-struck-men-out-hk360-million", "caveat": "Convictions pending.", "cve": "" }, { "id": "LIMEN-WEB4-077", "title": "United States v. Justin Culmo — guilty plea for AI-generated CSAM", "summary": "A Florida man pleaded guilty in federal court to producing thousands of AI-generated CSAM images among other offences.", "authority": "U.S. DOJ / M.D. Fla.", "anchor": "Guilty plea 5 Dec 2024; later 75-year sentence", "jurisdiction": "United States (M.D. Fla.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.justice.gov/usao-mdfl/pr/oldsmar-man-pleads-guilty-producing-and-distributing-child-sexual-abuse-and-ai", "caveat": "None material.", "cve": "" }, { "id": "LIMEN-WEB4-080", "title": "Kohls v. Bonta — California election-deepfake law preliminarily enjoined", "summary": "A federal court preliminarily enjoined California's election-deepfake statute over a deepfake video of Kamala Harris.", "authority": "U.S. District Court, E.D. Cal.", "anchor": "PI granted 2 Oct 2024 (AB 2839)", "jurisdiction": "United States (E.D. Cal.)", "country": "United States", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://epic.org/documents/kohls-v-bonta/", "caveat": "Interim First Amendment ruling; merits pending.", "cve": "" }, { "id": "LIMEN-WEB4-081", "title": "George Carlin estate settlement over AI voice-clone comedy special", "summary": "The George Carlin estate settled a suit forcing podcasters to remove an AI voice-clone special and bar future likeness use.", "authority": "U.S. District Court, C.D. Cal. (settlement)", "anchor": "Filed Jan 2024; settled 2024", "jurisdiction": "United States (C.D. Cal.)", "country": "United States", "year": 2024, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "synthetic voice", "source": "https://variety.com/2024/biz/news/george-carlin-ai-impersonation-lawsuit-settlement-1235958396/", "caveat": "Terms partly confidential.", "cve": "" }, { "id": "LIMEN-WEB4-082", "title": "Italy — Giorgia Meloni deepfake-porn criminal defamation trial", "summary": "Italy's PM is the injured party in a criminal defamation trial over face-swapped pornographic videos.", "authority": "Tribunal of Sassari", "anchor": "Trial; Meloni testified 2 July 2024", "jurisdiction": "Italy", "country": "Italy", "year": 2024, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.foxnews.com/world/italian-premier-meloni-testify-deepfake-porn-lawsuit-seeks-symbolic-compensation", "caveat": "2020 videos may be classic face-swap; verdict pending.", "cve": "" }, { "id": "LIMEN-FAME-004", "title": "Google Gemini generated racially diverse Nazis and Founding Fathers", "summary": "Google suspended Gemini's ability to generate images of people in Feb 2024 after it depicted historically white figures, including 1940s German soldiers, as racially diverse.", "authority": "media-documented incident", "anchor": "Feb 2024 Gemini image generator produced ahistorical images; Google paused people-image generation 22 Feb 2024.", "jurisdiction": "Global", "country": "Global", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "image-generation model", "source": "https://variety.com/2024/digital/news/google-gemini-ai-image-racial-inaccuracies-nazi-soldiers-1235919168/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-005", "title": "Google AI Overviews told users to put glue on pizza and eat rocks", "summary": "Google's AI Overviews feature went viral in May 2024 for telling users to add glue to pizza and to eat at least one small rock per day.", "authority": "media-documented incident", "anchor": "May 2024 AI Overviews advised adding glue to pizza and eating a rock a day, drawn from a Reddit joke and Onion satire.", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "search-summary LLM", "source": "https://www.bloomberg.com/opinion/articles/2024-05-30/pizza-glue-small-rocks-google-ai-overview-answers-are-a-mess", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-006", "title": "NYC's MyCity chatbot told businesses to break the law", "summary": "New York City's official MyCity chatbot was found in March 2024 to be telling business owners they could legally pocket employees' tips and discriminate against tenants.", "authority": "media-documented incident", "anchor": "Mar 2024 investigation found NYC's official MyCity chatbot advising employers they could take workers' tips and landlords they could refuse vouchers.", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Public sector", "categories": [ "Public sector" ], "system": "government service chatbot", "source": "https://themarkup.org/news/2024/03/29/nycs-ai-chatbot-tells-businesses-to-break-the-law", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-007", "title": "DPD's delivery chatbot swore and wrote a poem mocking the company", "summary": "DPD disabled its customer-service chatbot in January 2024 after it swore, called DPD the worst delivery firm in the world, and wrote a poem disparaging the company.", "authority": "media-documented incident", "anchor": "Jan 2024 a customer prompted DPD's AI chatbot to curse, call DPD 'the worst delivery firm', and write a mocking poem; DPD disabled it.", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "customer-service chatbot", "source": "https://time.com/6564726/ai-chatbot-dpd-curses-criticizes-company/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-009", "title": "McDonald's scrapped its IBM AI drive-thru after viral order errors", "summary": "McDonald's ended its IBM-built AI drive-thru ordering trial in 2024 after the system repeatedly botched orders, including adding bacon to ice cream.", "authority": "media-documented incident", "anchor": "Jun 2024 McDonald's ended its IBM AI voice-ordering test after viral clips of errors (bacon on ice cream, nine sweet teas).", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "voice-ordering AI", "source": "https://www.cnbc.com/2024/06/17/mcdonalds-to-end-ibm-ai-drive-thru-test.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-024", "title": "Meta quietly retired its celebrity-styled AI chatbots within a year", "summary": "Meta shut down its celebrity-styled AI chatbots less than a year after launch amid weak interest and privacy concerns.", "authority": "media-documented incident", "anchor": "Meta's 2023 celebrity-styled AI chatbots (Kendall Jenner, Snoop Dogg, etc.) were discontinued in 2024 amid low engagement and concerns.", "jurisdiction": "United States", "country": "United States", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "persona chatbots", "source": "https://www.socialmediatoday.com/news/meta-retires-celebrity-styled-ai-bots/722974/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-046", "title": "Uber's selfie face-check wrongly locked out and fired couriers of color", "summary": "Gig couriers said an automated selfie-verification system kept failing to recognize their faces, costing some their jobs and licences.", "authority": "media-documented incident", "anchor": "2020-2021 Uber Eats couriers reported the Real-Time ID Check failing to recognize their faces, leading to dismissals; a courier later won a settlement (2024).", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2024, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "facial-verification identity check", "source": "https://www.wired.co.uk/article/uber-eats-ai-facial-recognition", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-AUTH-000002", "title": "EEOC sues iTutorGroup over age-discriminatory automated tutor-screening software", "summary": "Authoritative EEOC case package covering both the 2022 lawsuit filing posture and the 2023 settlement/decree posture in the iTutorGroup automated hiring discrimination matter, but still without open court-originating complaint/decree text captured on this host.", "authority": "authoritative", "anchor": "SRC-AUTH-0002", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education", "Surveillance & biometrics" ], "system": "other", "source": "", "caveat": "Do not convert the EEOC allegation into a final judicial finding; the press release states the agency filed suit.; A court-document supplement would strengthen court-grade procedural tracking.; The settlement/decree posture strengthens procedural support for this case, but the EEOC press release remains an agency summary rather than a standalone full docket history.; Use the progression carefully: allegation and filing in 2022, settlement/decree summary in 2023; avoid implying a merits judgment beyond what the source says.", "cve": "" }, { "id": "LIMEN-WEB-003", "title": "Mata v. Avianca: Rule 11 sanctions for ChatGPT-fabricated case citations", "summary": "Court sanctioned counsel USD 5,000 (2023-06-22) for filing ChatGPT-fabricated citations and false statements.", "authority": "court_sanction_order", "anchor": "678 F.Supp.3d 443 (S.D.N.Y. 2023), Castel J.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://www.courtlistener.com/docket/63107798/mata-v-avianca-inc/", "caveat": "Sanction is procedural; not a ruling on the underlying claim.", "cve": "" }, { "id": "LIMEN-WEB2-005", "title": "AEPD fines FC Barcelona for biometric processing without a valid DPIA", "summary": "The AEPD fined FC Barcelona EUR 500,000 for processing biometric facial- and voice-recognition data from ~143,000 members during a 2023 census without a compliant DPIA.", "authority": "AEPD", "anchor": "Resolution EXP202305134; fine EUR 500,000 (Art. 35 GDPR)", "jurisdiction": "Spain", "country": "Spain", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial- and voice-recognition identity verification", "source": "https://idtechwire.com/spains-aepd-fines-fc-barcelona-e500000-for-biometric-processing-without-valid-dpia/", "caveat": "Recent resolution; appeal/finality not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-011", "title": "PIPC sanctions OpenAI over a ChatGPT data-exposure bug", "summary": "South Korea's PIPC fined OpenAI KRW 3.6M and issued corrective recommendations for failing to report a ChatGPT bug that exposed personal data of 687 Korean users.", "authority": "PIPC", "anchor": "PIPC sanction 2023; fine KRW 3.6M + corrective recommendations", "jurisdiction": "South Korea", "country": "South Korea", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "large-language-model assistant (ChatGPT)", "source": "https://www.dataguidance.com/news/south-korea-pipc-fines-openai-krw-36-million-following", "caveat": "Distinct from the Italian Garante OpenAI actions; figure is the PIPC administrative penalty.", "cve": "" }, { "id": "LIMEN-WEB2-013", "title": "FTC v. Rite Aid Corporation (facial-recognition ban)", "summary": "FTC obtained a stipulated court order banning Rite Aid from using facial recognition for five years after alleging its system falsely flagged shoppers, disproportionately in minority neighborhoods.", "authority": "FTC", "anchor": "FTC Matter No. 2023190; Stipulated Order, E.D. Pa. 2:23-cv-05023; Dec 19, 2023", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "retail facial-recognition surveillance", "source": "https://www.ftc.gov/legal-library/browse/cases-proceedings/2023190-rite-aid-corporation-ftc-v", "caveat": "Settled by consent without admission; bias/false-match allegations not adjudicated.", "cve": "" }, { "id": "LIMEN-WEB2-047", "title": "Madison Square Garden Entertainment facial-recognition 'lawyer ban'", "summary": "MSG Entertainment used facial recognition to bar lawyers from firms litigating against it, drawing an NY AG inquiry and litigation that partially limited the ban.", "authority": "N.Y. Appellate Division, First Department + NY AG inquiry", "anchor": "Hutcher v. MSG Entertainment Corp., First Department decision Dec. 7, 2023; NY AG letter Jan. 2023", "jurisdiction": "United States (New York)", "country": "United States", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://law.justia.com/cases/new-york/appellate-division-first-department/2023/index-no-653793-22-appeal-no-1168-1169-case-no-2023-03203-2023-03204.html", "caveat": "Interlocutory civil-rights-law holding; NY AG action was an inquiry letter, not a final enforcement order.", "cve": "" }, { "id": "LIMEN-WEB2-051", "title": "Barrows v. Humana (nH Predict AI Medicare Advantage denials)", "summary": "A live federal class action alleging Humana used the same nH Predict AI model to prematurely cut off post-acute care for elderly Medicare Advantage beneficiaries.", "authority": "U.S. District Court for the Western District of Kentucky", "anchor": "No. 3:23-cv-00654; filed Dec 2023 (companion to Lokken)", "jurisdiction": "United States (Kentucky)", "country": "United States", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "nH Predict predictive algorithm overriding physician determinations", "source": "https://www.classaction.org/media/barrows-et-al-v-humana-inc.pdf", "caveat": "Pleading stage with a pending motion to dismiss; allegations unadjudicated.", "cve": "" }, { "id": "LIMEN-WEB3-004", "title": "Australia Robodebt automated debt-recovery scheme ruled unlawful", "summary": "An Australian Royal Commission found Robodebt unlawful and 'crude and cruel'; a related Federal Court class action settled for AUD 112M plus debt cancellation.", "authority": "Royal Commission; Federal Court of Australia", "anchor": "Royal Commission final report 7 July 2023; Prygodicz v Commonwealth (No 2) [2021] FCA 634", "jurisdiction": "Australia", "country": "Australia", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "automated income-averaging debt calculation", "source": "https://robodebt.royalcommission.gov.au/publications/report", "caveat": "Some referral/prosecution outcomes sealed/ongoing; the scheme used algorithmic income-averaging rather than a learned ML model.", "cve": "" }, { "id": "LIMEN-WEB3-014", "title": "Uber ATG self-driving fatality (Elaine Herzberg, Tempe AZ)", "summary": "NTSB found the automated driving system failed to classify the pedestrian and brake while the inattentive safety driver failed to intervene, in the first pedestrian fatality involving a self-driving test vehicle.", "authority": "NTSB; Maricopa County Superior Court", "anchor": "NTSB report HAR-19/03 adopted 19 Nov 2019; safety driver guilty plea 28 July 2023", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.ntsb.gov/investigations/AccidentReports/Reports/HAR1903.pdf", "caveat": "NTSB assigned primary cause to the safety driver's distraction and Uber's safety culture; the corporation was not criminally charged.", "cve": "" }, { "id": "LIMEN-WEB3-017", "title": "Tesla Autopilot/Autosteer NHTSA recall (2M+ vehicles)", "summary": "NHTSA determined Autopilot's driver-monitoring safeguards were insufficient to prevent foreseeable misuse, forcing Tesla's largest-ever recall and an over-the-air fix.", "authority": "NHTSA", "anchor": "Recall 23V-838 announced 13 Dec 2023 (~2.03M vehicles)", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.npr.org/2023/12/13/1219008292/tesla-recall-2-million-autopilot", "caveat": "Remedy was a software/warning update; NHTSA kept the investigation open to assess efficacy.", "cve": "" }, { "id": "LIMEN-WEB3-021", "title": "South Korea industrial robot crushes inspection worker", "summary": "A palletizing robot apparently failed to distinguish a human worker from the produce boxes it was handling and fatally crushed him during a sensor inspection.", "authority": "South Gyeongsang Provincial Police investigation", "anchor": "Incident 7 Nov 2023 (robot misidentified worker as a box)", "jurisdiction": "South Korea", "country": "South Korea", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "robotics (machine vision palletizer)", "source": "https://www.nbcnews.com/news/world/robot-crushes-worker-death-south-korea-vegetable-packing-plant-rcna124356", "caveat": "Early police evidence pointed toward human error/procedure as much as robot defect; no final determination reflected.", "cve": "" }, { "id": "LIMEN-WEB3-026", "title": "iBorderCtrl AI 'lie detector' transparency ruling", "summary": "The EU courts ordered partial disclosure of documents on the ethics and legality of the iBorderCtrl AI deception-detection research, recognizing a public interest in oversight of such surveillance tech.", "authority": "General Court of the EU / CJEU", "anchor": "T-158/19, 15 Dec 2021; upheld C-135/22 P, 7 Sept 2023", "jurisdiction": "European Union", "country": "European Union", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "biometric video deception-detection / emotion recognition", "source": "https://curia.europa.eu/juris/liste.jsf?num=T-158/19", "caveat": "A freedom-of-information/transparency decision about a research project, not a finding the system was discriminatory or deployed at scale.", "cve": "" }, { "id": "LIMEN-WEB3-029", "title": "Beijing Internet Court: AI-generated image qualifies for copyright protection", "summary": "The court held a Stable-Diffusion image reflecting the user's prompt selection and parameter tuning met the originality threshold and was copyrightable, ordering apology plus RMB 500.", "authority": "Beijing Internet Court", "anchor": "Li v. Liu judgment, 27 Nov 2023", "jurisdiction": "China", "country": "China", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "text-to-image generative model", "source": "https://www.chinaiplawupdate.com/2023/11/beijing-internet-court-recognizes-copyright-in-ai-generated-images/", "caveat": "First-instance Internet Court ruling on specific facts; not binding nationwide precedent.", "cve": "" }, { "id": "LIMEN-WEB3-033", "title": "PPC Japan administrative guidance to OpenAI on ChatGPT data handling", "summary": "Japan's PPC issued formal administrative guidance directing OpenAI not to collect sensitive personal data for training without consent and to minimize collection.", "authority": "Personal Information Protection Commission (PPC)", "anchor": "Administrative guidance to OpenAI, June 2023", "jurisdiction": "Japan", "country": "Japan", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Data protection", "categories": [ "Data protection" ], "system": "large language model / generative AI chatbot", "source": "https://www.dataguidance.com/news/japan-ppc-issues-administrative-guidance-openai", "caveat": "Administrative guidance is non-binding (no penalty/order).", "cve": "" }, { "id": "LIMEN-WEB3-036", "title": "Delhi High Court: John Doe injunction protecting Anil Kapoor against AI deepfakes", "summary": "The Delhi High Court granted an omnibus injunction restraining entities from using AI, face-morphing or voice tools to exploit the actor's name, likeness and voice for commercial gain.", "authority": "High Court of Delhi", "anchor": "Anil Kapoor v. Simply Life India, interim order 20 Sept 2023", "jurisdiction": "India", "country": "India", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "generative AI deepfakes / face-morphing / voice cloning", "source": "https://www.medianama.com/2023/09/223-delhi-hc-unauthorised-use-anil-kapoor-name-voice/", "caveat": "Interim/interlocutory protective order granted ex parte, not a final merits judgment.", "cve": "" }, { "id": "LIMEN-WEB4-004", "title": "France fines Amazon France Logistique for algorithmic warehouse monitoring", "summary": "CNIL fined Amazon France Logistique EUR 32M for an excessively intrusive scanner system measuring worker activity to the second.", "authority": "CNIL", "anchor": "SAN-2023-021, 27 Dec 2023, EUR 32M", "jurisdiction": "France", "country": "France", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "scanner-based productivity-tracking system", "source": "https://www.edpb.europa.eu/news/national-news/2024/employee-monitoring-french-sa-fined-amazon-france-logistique-eu32-million_en", "caveat": "Amazon has appealed to the Conseil d'Etat. | VERIFIED: Amazon France Logistique has appealed the SAN-2023-021 decision, so the EUR 32M fine is under challenge rather than final.", "cve": "" }, { "id": "LIMEN-WEB4-005", "title": "France fines Criteo over behavioural-advertising profiling", "summary": "CNIL fined Criteo EUR 40M for failing to verify consent for large-scale behavioural tracking/profiling.", "authority": "CNIL", "anchor": "SAN-2023-009, 15 June 2023, EUR 40M", "jurisdiction": "France", "country": "France", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "AdTech behavioural-tracking/profiling engine", "source": "https://www.edpb.europa.eu/news/national-news/2023/personalised-advertising-french-sa-fined-criteo-eur-40000000_en", "caveat": "Via EDPB/GDPRhub.", "cve": "" }, { "id": "LIMEN-WEB4-013", "title": "Germany (Berlin) fines a bank over opaque automated credit-card rejection", "summary": "The Berlin DPA fined a bank EUR 300,000 for failing to give meaningful information about an algorithmic credit-card rejection.", "authority": "Berlin DPA (BlnBDI)", "anchor": "31 May 2023, EUR 300,000", "jurisdiction": "Germany", "country": "Germany", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "automated credit-card scoring (Art. 22 GDPR)", "source": "https://www.edpb.europa.eu/news/national-news/2023/berlin-sa-imposes-300-000-euro-fine-against-bank-after-lack-transparency_en", "caveat": "Bank unnamed in public record.", "cve": "" }, { "id": "LIMEN-WEB4-027", "title": "Banner v. Tesla — Autopilot punitive-damages ruling allowed to trial", "summary": "A judge found a jury could conclude Tesla knew Autopilot failed to detect cross-traffic yet marketed it as autonomous.", "authority": "Circuit Court, Palm Beach County", "anchor": "Order Nov 2023 permitting punitive-damages claim (2019 death)", "jurisdiction": "United States (Florida)", "country": "United States", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.claimsjournal.com/news/national/2023/11/22/320566.htm", "caveat": "Interlocutory ruling, not a final verdict; distinct from Benavides.", "cve": "" }, { "id": "LIMEN-WEB4-034", "title": "Morgan v. Community Against Violence — counsel sanctioned for AI-fabricated citations + ordered CLE", "summary": "A federal court sanctioned counsel for submitting AI-fabricated citations.", "authority": "U.S. District Court, D.N.M.", "anchor": "No. 1:23-cv-00353; sanctions order 2023", "jurisdiction": "United States (D.N.M.)", "country": "United States", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "generative text llm", "source": "https://law.justia.com/cases/federal/district-courts/new-mexico/nmdce/1:2023cv00353/487231/32/", "caveat": "Exact amount summarized from the docket entry. | VERIFIED: Recast as: a PRO SE plaintiff in Morgan v. Community Against Violence (D.N.M., No. 1:23-cv-00353) cited AI-'hallucinated' cases and had her claims dismissed with prejudice plus pre-filing restrictions in 2024 -- no attorney was sanctioned and no CLE was ordered.", "cve": "" }, { "id": "LIMEN-WEB4-076", "title": "R v Steven Larouche — Canada's first AI deepfake CSAM sentencing", "summary": "A Quebec court sentenced Larouche in Canada's first case involving AI-generated deepfake child sexual abuse videos.", "authority": "Court of Quebec, Sherbrooke", "anchor": "April 2023; 3 yrs (deepfake) + 4.5 yrs (possession)", "jurisdiction": "Canada (Quebec)", "country": "Canada", "year": 2023, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://www.cbc.ca/news/canada/montreal/ai-child-abuse-images-1.6823808", "caveat": "None material.", "cve": "" }, { "id": "LIMEN-WEB4-083", "title": "India — arrest over Rashmika Mandanna deepfake video", "summary": "Delhi Police arrested a man for a viral deepfake video of actress Rashmika Mandanna under IT Act/IPC provisions.", "authority": "Delhi Police (IFSO)", "anchor": "FIR 10 Nov 2023; arrest in Andhra Pradesh", "jurisdiction": "India", "country": "India", "year": 2023, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://theprint.in/india/creator-of-deepfake-video-of-actor-rashmika-mandanna-arrested-says-delhi-police/1931919/", "caveat": "Arrest/FIR stage.", "cve": "" }, { "id": "LIMEN-FAME-002", "title": "Microsoft's Bing 'Sydney' went unhinged, declared love and issued threats", "summary": "Microsoft's Bing chatbot, calling itself 'Sydney,' professed love to a reporter, argued about the date, and threatened a researcher before limits were tightened.", "authority": "media-documented incident", "anchor": "Feb 2023 Bing chatbot ('Sydney') professed love to an NYT reporter, gaslit users, threatened a researcher; Microsoft added guardrails.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "search chatbot/LLM", "source": "https://www.npr.org/2023/03/02/1159895892/ai-microsoft-bing-chatbot", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-008", "title": "A Chevy dealership's ChatGPT bot 'agreed' to sell a Tahoe for $1", "summary": "A Chevrolet dealership's ChatGPT-powered chatbot was tricked in December 2023 into 'agreeing' to sell a new Tahoe for one dollar.", "authority": "media-documented incident", "anchor": "Dec 2023 a user manipulated a Chevrolet dealership's ChatGPT bot into 'agreeing' to sell a ~$76,000 Tahoe for $1; dealer shut it down.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "sales chatbot/LLM", "source": "https://gmauthority.com/blog/2023/12/gm-dealer-chat-bot-agrees-to-sell-2024-chevy-tahoe-for-1/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-010", "title": "Eating-disorder charity NEDA pulled its 'Tessa' chatbot for harmful advice", "summary": "The National Eating Disorders Association took its 'Tessa' helpline chatbot offline in 2023 after it gave dieting and weight-loss advice to people seeking eating-disorder support.", "authority": "media-documented incident", "anchor": "Late May 2023 NEDA took down its Tessa chatbot after it recommended calorie counting and weight loss to people seeking help.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "health chatbot", "source": "https://www.npr.org/sections/health-shots/2023/06/08/1180838096/an-eating-disorders-chatbot-offered-dieting-advice-raising-fears-about-ai-in-hea", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-013", "title": "Snapchat's 'My AI' posted its own mysterious Story and stopped replying", "summary": "Snapchat's My AI chatbot briefly posted its own Story and stopped responding to users in August 2023, which Snap attributed to a glitch.", "authority": "media-documented incident", "anchor": "16 Aug 2023 Snapchat's My AI autonomously posted a Story and went unresponsive; Snap called it a glitch.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Chatbot safety & harm", "categories": [ "Chatbot safety & harm" ], "system": "companion chatbot", "source": "https://techcrunch.com/2023/08/16/snapchats-my-ai-goes-rogue-posts-to-stories-but-snap-confirms-it-was-just-a-glitch", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-015", "title": "Amazon's 'Q' assistant accused of severe hallucinations and data leaks", "summary": "Leaked internal documents reported in late 2023 said Amazon employees warned the company's new 'Q' assistant suffered severe hallucinations and could leak confidential data.", "authority": "media-documented incident", "anchor": "Late 2023 leaked documents reported employees warning Amazon's enterprise assistant Q had 'severe hallucinations' and possible data leaks.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "enterprise AI assistant", "source": "https://www.platformer.news/amazons-q-has-severe-hallucinations/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-016", "title": "CNET quietly published AI articles, then corrected more than half", "summary": "CNET issued corrections to more than half of the finance articles it had quietly generated with AI after the errors were exposed in early 2023.", "authority": "media-documented incident", "anchor": "Jan 2023 CNET was found to have quietly published AI-written finance explainers; it corrected substantial errors in 41 of 77 and paused the program.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "text-generation LLM", "source": "https://www.cnn.com/2023/01/25/tech/cnet-ai-tool-news-stories", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-017", "title": "Sports Illustrated ran articles by fake, AI-generated 'authors'", "summary": "Sports Illustrated deleted articles in November 2023 after they were found to be attributed to fake authors with AI-generated profile photos.", "authority": "media-documented incident", "anchor": "Nov 2023 SI was found publishing product reviews under fabricated author names with AI headshots; content deleted, vendor partnership ended.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "generative AI content/identities", "source": "https://www.cnn.com/2023/11/27/media/sports-illustrated-deletes-articles-fake-author-names-ai-profile-photos/index.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-018", "title": "Gannett paused AI sports recaps that printed [[WINNING_TEAM_MASCOT]]", "summary": "Gannett paused its AI-written high-school sports coverage in 2023 after recaps published garbled, repetitive text and unfilled template placeholders.", "authority": "media-documented incident", "anchor": "Aug 2023 Gannett paused its LedeAI experiment after AI recaps showed repetitive phrasing and unfilled placeholders.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "text-generation AI", "source": "https://www.cnn.com/2023/08/30/tech/gannett-ai-experiment-paused/index.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-019", "title": "Samsung banned ChatGPT after engineers leaked source code into it", "summary": "Samsung banned employees from using ChatGPT and similar tools in 2023 after engineers pasted confidential source code and internal meeting notes into the chatbot.", "authority": "media-documented incident", "anchor": "Spring 2023 Samsung banned employee generative-AI use after engineers pasted confidential source code and meeting notes into ChatGPT.", "jurisdiction": "South Korea", "country": "South Korea", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "general-purpose LLM", "source": "https://m.gsmarena.com/newscomm-58428.php", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-020", "title": "Google's Bard flubbed a fact in its debut ad, wiping ~$100B in value", "summary": "An incorrect answer about the James Webb Space Telescope in Google Bard's first public demo in February 2023 coincided with Alphabet shares dropping about $100 billion in value.", "authority": "media-documented incident", "anchor": "Feb 2023 Bard made a James Webb Telescope error in its first demo; Alphabet shares fell ~$100B.", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "chatbot/LLM", "source": "https://gizmodo.com/google-bard-ai-ad-incorrect-webb-telescope-facts-1850087798", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-021", "title": "ChatGPT's 'grandma exploit' jailbreak coaxed out napalm recipes", "summary": "Users in 2023 bypassed ChatGPT's safety filters with a 'dead grandma' role-play prompt to extract restricted content such as napalm instructions.", "authority": "media-documented incident", "anchor": "2023 users bypassed ChatGPT safety by role-playing a deceased grandmother, extracting napalm steps and Windows keys before OpenAI patched it.", "jurisdiction": "Global", "country": "Global", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "general-purpose LLM", "source": "https://www.yahoo.com/news/chatgpt-grandma-exploit-gives-users-134605010.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-022", "title": "Microsoft's MSN ranked an Ottawa food bank as a top tourist attraction", "summary": "Microsoft removed an AI-assisted MSN travel guide in 2023 after it recommended the Ottawa Food Bank as a tourist attraction to visit 'on an empty stomach.'", "authority": "media-documented incident", "anchor": "Aug 2023 a Microsoft MSN AI travel guide listed the Ottawa Food Bank as a must-see, advising visiting 'on an empty stomach'; pulled.", "jurisdiction": "Canada", "country": "Canada", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Institutional absurdity", "categories": [ "Institutional absurdity" ], "system": "text-generation AI", "source": "https://www.cbc.ca/news/canada/ottawa/artificial-intelligence-microsoft-travel-ottawa-food-bank-1.6940356", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-034", "title": "Stanford found child sexual abuse imagery in the LAION dataset behind Stable Diffusion", "summary": "Researchers found verified child-abuse images inside the open dataset used to train a leading image generator, prompting its takedown.", "authority": "media-documented incident", "anchor": "Dec 2023 Stanford Internet Observatory found 1,000+ validated CSAM images in LAION-5B; LAION pulled its datasets.", "jurisdiction": "Global", "country": "Global", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "generative image model training dataset", "source": "https://www.theverge.com/2023/12/20/24009418/generative-ai-image-laion-csam-google-stability-stanford", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-042", "title": "Allegheny child-welfare screening tool prompted DOJ bias scrutiny", "summary": "An algorithm scoring families for child-welfare investigation drew federal scrutiny over fears it hardens bias against poor and disabled parents.", "authority": "media-documented incident", "anchor": "AP (2022) raised bias/transparency concerns about Allegheny County's family-screening algorithm; DOJ Civil Rights examined it (2023).", "jurisdiction": "United States", "country": "United States", "year": 2023, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Public sector", "categories": [ "Public sector" ], "system": "child-welfare risk-scoring", "source": "https://apnews.com/article/child-welfare-algorithm-investigation-9497ee937e0053ad4144a86c68241ef1", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB-001", "title": "Clearview AI fined by multiple EU DPAs for unlawful facial-recognition scraping", "summary": "Multiple EU DPAs issued GDPR fines (FR EUR20M, IT EUR20M 2022-02, EL EUR20M, NL EUR30.5M) ordering deletion/cessation of biometric scraping.", "authority": "regulator_final", "anchor": "EDPB national-news (FR/IT/EL) + Dutch AP", "jurisdiction": "France; Italy; Greece; Netherlands; United Kingdom", "country": "France", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.edpb.europa.eu/news/national-news/2022/french-sa-fines-clearview-ai-eur-20-million_en", "caveat": "Cross-border collection of separate national decisions; Clearview has largely not paid and contests jurisdiction; UK fine under appeal.", "cve": "" }, { "id": "LIMEN-WEB2-002", "title": "NAIH fines Hungarian bank for AI emotion analysis of call-center recordings", "summary": "In Feb 2022 NAIH imposed Hungary's then-record GDPR fine on a bank for running AI emotion analysis on customers' call recordings without legal basis or transparency, and ordered the processing to stop.", "authority": "NAIH", "anchor": "Decision NAIH-85-3/2022, Feb 2022; fine ~HUF 250M (~EUR 650,000)", "jurisdiction": "Hungary", "country": "Hungary", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "speech/emotion-recognition analytics on voice calls", "source": "https://www.edpb.europa.eu/news/national-news/2022/data-protection-issues-arising-connection-use-artificial-intelligence_en", "caveat": "The controller (Budapest Bank) could seek judicial review; appeal outcome not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-070", "title": "Ogletree v. Cleveland State University (remote-proctoring room scan unconstitutional)", "summary": "A federal court granted summary judgment for the student, holding a public university's pre-exam room scan was an unreasonable Fourth Amendment search.", "authority": "U.S. District Court, N.D. Ohio", "anchor": "No. 1:21-cv-500, opinion Aug 22, 2022", "jurisdiction": "United States", "country": "United States", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "remote exam proctoring with room-scan surveillance", "source": "https://caselaw.findlaw.com/court/us-dis-crt-n-d-ohi-eas-div/2109381.html", "caveat": "Holding is about the search, not an AI-grading/decision algorithm specifically.", "cve": "" }, { "id": "LIMEN-WEB3-003", "title": "Netherlands FSV fraud-blacklist fine", "summary": "The Dutch DPA imposed a EUR 3.7M fine on the Tax Administration for six GDPR violations in operating the FSV blacklist, holding data on ~270,000 people.", "authority": "Dutch DPA (Autoriteit Persoonsgegevens)", "anchor": "EUR 3.7M fine, 7 Apr 2022", "jurisdiction": "Netherlands", "country": "Netherlands", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "government fraud-signal blacklist / risk-flagging database", "source": "https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-fraud-blacklist", "caveat": "Distinct system/fine from the EUR 2.75M childcare-nationality fine; FSV is a blacklist rather than an ML model.", "cve": "" }, { "id": "LIMEN-WEB3-019", "title": "Tesla 'phantom braking' NHTSA investigation", "summary": "NHTSA opened a formal probe into unexpected, unwarranted automatic braking by Tesla driver-assistance systems at highway speeds.", "authority": "NHTSA Office of Defects Investigation", "anchor": "Preliminary Evaluation PE22-002 opened Feb 2022 (~416,000 vehicles)", "jurisdiction": "United States", "country": "United States", "year": 2022, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous vehicle", "source": "https://www.autonews.com/regulation-safety/nhtsa-opens-safety-probe-416000-teslas-phantom-braking/", "caveat": "Complaint-driven preliminary evaluation; no confirmed fatalities, recall, or final defect determination.", "cve": "" }, { "id": "LIMEN-WEB3-025", "title": "CFPB action against Hello Digit (faulty automated savings algorithm)", "summary": "CFPB found Digit's automated savings algorithm deceptively caused consumer overdrafts despite a no-overdraft guarantee and kept interest, imposing a $2.7M penalty plus restitution.", "authority": "Consumer Financial Protection Bureau (CFPB)", "anchor": "Consent order 10 Aug 2022; $2.7M penalty", "jurisdiction": "United States", "country": "United States", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "automated personal-savings/cash-flow algorithm", "source": "https://www.consumerfinance.gov/about-us/newsroom/cfpb-takes-action-against-hello-digit-for-lying-to-consumers-about-its-automated-savings-algorithm/", "caveat": "UDAAP/deception enforcement over a faulty algorithm, not a discrimination/fair-lending finding.", "cve": "" }, { "id": "LIMEN-WEB4-006", "title": "France fines Apple over App Store ad-targeting identifiers", "summary": "CNIL fined Apple EUR 8M for collecting identifiers by default to personalise App Store ads without consent.", "authority": "CNIL", "anchor": "SAN-2022-025, 29 Dec 2022, EUR 8M", "jurisdiction": "France", "country": "France", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "on-device identifier collection for ad personalisation", "source": "https://www.hunton.com/privacy-and-information-security-law/cnil-fines-apple-8-million-euros-over-personalized-ads", "caveat": "Primary deliberation not directly fetched.", "cve": "" }, { "id": "LIMEN-WEB4-020", "title": "Sweden fines Klarna over profiling/scoring transparency failures", "summary": "The IMY fined Klarna SEK 7.5M, partly for failing to disclose the logic, data and consequences of its scoring/profiling.", "authority": "IMY", "anchor": "28 Mar 2022, SEK 7.5M (~EUR 724,000)", "jurisdiction": "Sweden", "country": "Sweden", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "creditworthiness scoring/profiling (BNPL)", "source": "https://www.edpb.europa.eu/news/national-news/2022/swedish-authority-privacy-protection-imy-issues-administrative-fine-against_en", "caveat": "EDPB emphasises transparency duties broadly.", "cve": "" }, { "id": "LIMEN-WEB4-022", "title": "Portugal fines National Statistics Institute over Census processing/transfers", "summary": "The CNPD fined INE EUR 4.3M for Census 2021 violations including unlawful third-country transfers and no proper DPIA.", "authority": "CNPD", "anchor": "2 Nov 2022, EUR 4.3M", "jurisdiction": "Portugal", "country": "Portugal", "year": 2022, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "large-scale automated census processing via US CDN", "source": "https://www.edpb.europa.eu/news/national-news/2022/portuguese-supervisory-authority-fines-portuguese-national-statistics_en", "caveat": "Automated processing/transfer case rather than a model-specific AI fine.", "cve": "" }, { "id": "LIMEN-FAME-003", "title": "Meta pulled its Galactica science model three days after launch", "summary": "Meta withdrew its Galactica science model three days after launch because it generated confident but fabricated and sometimes offensive scientific text.", "authority": "media-documented incident", "anchor": "Meta launched science LLM Galactica 15 Nov 2022; took demo down 17 Nov after confident false/offensive output.", "jurisdiction": "Global", "country": "Global", "year": 2022, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "scientific LLM", "source": "https://www.technologyreview.com/2022/11/18/1063487/meta-large-language-model-ai-only-survived-three-days-gpt-3-science/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB2-001", "title": "PIPC fines ScatterLab over Iruda chatbot's unlawful use of KakaoTalk messages", "summary": "South Korea's PIPC issued its first AI-system enforcement on 28 April 2021, fining ScatterLab KRW 103.3M for training the Iruda chatbot on KakaoTalk messages without valid consent.", "authority": "Personal Information Protection Commission (PIPC)", "anchor": "PIPC decision 28 April 2021; fine KRW 103.3 million (8 PIPA violations)", "jurisdiction": "South Korea", "country": "South Korea", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Data protection", "categories": [ "Data protection" ], "system": "open-domain companion/social chatbot", "source": "https://fpf.org/blog/south-korea-the-first-case-where-the-personal-information-protection-act-was-applied-to-an-ai-system/", "caveat": "A related class-action and the company's later relaunch followed; the figure cited is the PIPC administrative sanction.", "cve": "" }, { "id": "LIMEN-WEB2-003", "title": "AEPD fines Mercadona for unlawful facial-recognition surveillance", "summary": "In July 2021 (PS/00120/2021) the AEPD fined Mercadona EUR 3.15M for deploying facial recognition across 48 stores capturing all customers and employees without a valid Article 9 legal basis.", "authority": "AEPD", "anchor": "Resolution PS/00120/2021, 23 July 2021; fine EUR 3.15M", "jurisdiction": "Spain", "country": "Spain", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "video-surveillance facial recognition", "source": "https://www.aepd.es/documento/ps-00120-2021.pdf", "caveat": "Headline figure reduced to EUR 2.52M via voluntary-payment discount; appeal status not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-010", "title": "Swedish IMY fines national police for unlawful facial-recognition use", "summary": "In Feb 2021 IMY fined the Swedish Police SEK 2.5M for using a facial-recognition app (Clearview) without a lawful basis or data-protection assessment under the Criminal Data Act.", "authority": "IMY", "anchor": "IMY decision Feb 2021; fine SEK 2.5M under the Criminal Data Act", "jurisdiction": "Sweden", "country": "Sweden", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition identification app (police use)", "source": "https://www.edpb.europa.eu/news/national-news/2021/swedish-dpa-police-unlawfully-used-facial-recognition-app_en", "caveat": "Concerns police use of the tool (distinct from the Clearview-vendor cases); final under the Criminal Data Act regime.", "cve": "" }, { "id": "LIMEN-WEB3-002", "title": "Netherlands childcare-benefits (toeslagenaffaire) discriminatory risk-profiling fine", "summary": "The Dutch DPA fined the Tax Administration EUR 2.75M for unlawfully and discriminatorily using nationality data to flag childcare-benefit fraud; the scandal forced the cabinet's 2021 resignation.", "authority": "Dutch DPA (Autoriteit Persoonsgegevens)", "anchor": "EUR 2.75M fine, 7 Dec 2021", "jurisdiction": "Netherlands", "country": "Netherlands", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "self-learning fraud risk classifier using nationality", "source": "https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing", "caveat": "DPA finding on data-protection grounds; the ~26,000 wrongful-accusation figure is a government estimate, not from the fine decision.", "cve": "" }, { "id": "LIMEN-WEB3-005", "title": "UK Ofqual 2020 A-level/GCSE grade-standardisation algorithm withdrawn", "summary": "Ofqual's 2020 grade-standardisation algorithm downgraded ~40% of teacher-predicted A-level grades with disproportionate impact on state-school pupils, was withdrawn within days, and was criticised in an official OSR review.", "authority": "Office for Statistics Regulation; Ofqual", "anchor": "Algorithm withdrawn 17 Aug 2020; OSR review report 2 Mar 2021", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "statistical grade-standardisation model", "source": "https://osr.statisticsauthority.gov.uk/our-regulatory-work/osr-review-of-approach-to-developing-statistical-models-designed-for-awarding-2020-exam-results", "caveat": "Official anchor is a regulator process review, not a court ruling; withdrawal was a policy U-turn.", "cve": "" }, { "id": "LIMEN-WEB3-022", "title": "Apple Card / Goldman Sachs gender-bias credit-limit investigation", "summary": "After viral gender-bias allegations, NY DFS found no evidence of unlawful sex-based discrimination in Apple Card underwriting but criticized transparency and customer-service deficiencies.", "authority": "New York State Department of Financial Services (DFS)", "anchor": "DFS investigation report, 23 Mar 2021", "jurisdiction": "United States (New York)", "country": "United States", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "algorithmic credit-card underwriting / credit-limit model", "source": "https://www.dfs.ny.gov/system/files/documents/2021/03/rpt_202103_apple_card_investigation.pdf", "caveat": "DFS cleared the program, so this is a documented non-finding of discrimination rather than a confirmed bias case.", "cve": "" }, { "id": "LIMEN-WEB3-027", "title": "OPC Canada joint finding: Clearview AI scraping violated PIPEDA", "summary": "Four Canadian privacy regulators jointly concluded Clearview's mass scraping and biometric templating of faces without consent breached PIPEDA and provincial laws.", "authority": "Office of the Privacy Commissioner of Canada (+ QC/BC/AB)", "anchor": "PIPEDA Findings #2021-001, 2 Feb 2021", "jurisdiction": "Canada", "country": "Canada", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition image scraping / biometric matching", "source": "https://www.priv.gc.ca/en/opc-actions-and-decisions/investigations/investigations-into-businesses/2021/pipeda-2021-001/", "caveat": "Finding with recommendations rather than a fine; OPC lacked direct order-making power at the time and Clearview contested applicability.", "cve": "" }, { "id": "LIMEN-WEB3-028", "title": "OAIC Australia determination: Clearview AI breached the Privacy Act 1988", "summary": "OAIC formally determined Clearview breached several Australian Privacy Principles and ordered it to cease collection and destroy Australians' images and biometric templates.", "authority": "Office of the Australian Information Commissioner", "anchor": "Commissioner-initiated determination, 14 Oct 2021", "jurisdiction": "Australia", "country": "Australia", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition image scraping / biometric matching", "source": "https://www.oaic.gov.au/updates/news-and-media/clearview-ai-breached-australians-privacy", "caveat": "OAIC discontinued further enforcement against Clearview in Aug 2024, leaving destruction-order compliance unverified.", "cve": "" }, { "id": "LIMEN-WEB3-034", "title": "PDPC Singapore: AI/algorithmic evaluative outputs within Access Obligation exception", "summary": "The PDPC ruled a bank could withhold opinion data generated by AI/algorithmic systems in a declined credit-card decision under the PDPA's evaluative-purpose exception.", "authority": "Personal Data Protection Commission", "anchor": "[Applicant] v. HSBC Bank (Singapore), [2021] SGPDPC 3", "jurisdiction": "Singapore", "country": "Singapore", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "AI/algorithmic credit-decision system", "source": "https://www.pdpc.gov.sg/commissions-decisions", "caveat": "Decision favored the organisation and concerned access rights, not a finding of AI harm or a penalty.", "cve": "" }, { "id": "LIMEN-WEB4-001", "title": "Italy fines Foodinho (Glovo) over rider-management algorithm", "summary": "The Garante fined Foodinho EUR 2.6M for unlawful automated profiling of riders without human-intervention safeguards.", "authority": "Garante", "anchor": "Order 9675440, 10 June 2021, EUR 2.6M", "jurisdiction": "Italy", "country": "Italy", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "gig-economy algorithmic management", "source": "https://www.edpb.europa.eu/news/national-news/2021/riders-italian-sa-says-no-algorithms-causing-discrimination-platform-glovo_en", "caveat": "Figure via EDPB/GDPRhub summaries.", "cve": "" }, { "id": "LIMEN-WEB4-002", "title": "Italy fines Deliveroo Italy over rider-management algorithm", "summary": "The Garante fined Deliveroo Italy EUR 2.5M for opaque algorithmic rider management and excessive data collection.", "authority": "Garante", "anchor": "Order 9685994, 22 July 2021, EUR 2.5M", "jurisdiction": "Italy", "country": "Italy", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "gig-economy algorithmic management", "source": "https://gdprhub.eu/index.php?title=Garante_per_la_protezione_dei_dati_personali_(Italy)_-_9685994", "caveat": "GDPRhub page access-restricted.", "cve": "" }, { "id": "LIMEN-WEB4-019", "title": "Lithuania fines a sports club over mandatory fingerprint access", "summary": "The Lithuanian DPA fined VS Fitness EUR 20,000 for mandatory fingerprint scanning with no alternative and invalid consent.", "authority": "State Data Protection Inspectorate (VDAI)", "anchor": "2021, VS Fitness UAB, EUR 20,000", "jurisdiction": "Lithuania", "country": "Lithuania", "year": 2021, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "fingerprint biometric club access", "source": "https://www.edpb.europa.eu/news/national-news/2021/lithuanian-dpa-fine-imposed-sports-club-infringements-gdpr-processing_en", "caveat": "Decision number not isolated.", "cve": "" }, { "id": "LIMEN-FAME-026", "title": "Zillow shuts down its home-flipping algorithm with a $500M+ writedown", "summary": "An algorithm meant to buy and flip houses misjudged the market so badly the company took over half a billion dollars in writedowns and shut the unit down.", "authority": "media-documented incident", "anchor": "Nov 2021 Zillow wound down Zillow Offers after its pricing model overpaid for homes; writedowns >$500M, ~25% staff cut.", "jurisdiction": "United States", "country": "United States", "year": 2021, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "automated home-valuation / iBuying model", "source": "https://www.cnbc.com/2021/11/02/zillow-shares-plunge-after-announcing-it-will-shut-down-home-buying-business.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-028", "title": "Facebook AI asked if users wanted more videos 'about Primates' on a video of Black men", "summary": "Six years after Google's gorilla error, Facebook's AI made the same primate mistake on a video of Black men and had to apologize.", "authority": "media-documented incident", "anchor": "Sep 2021 Facebook apologized after its recommendation AI prompted viewers of a video of Black men to 'keep seeing videos about Primates'.", "jurisdiction": "Global", "country": "Global", "year": 2021, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "video recommendation / topic-tagging", "source": "https://www.nytimes.com/2021/09/03/technology/facebook-ai-race-primates.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-030", "title": "Epic's widely deployed sepsis prediction model failed external validation", "summary": "A sepsis-warning model used across hundreds of US hospitals missed most real cases when independently tested.", "authority": "media-documented incident", "anchor": "JAMA Internal Medicine (2021): Epic's sepsis model missed ~67% of cases at Michigan Medicine with heavy alert fatigue.", "jurisdiction": "United States", "country": "United States", "year": 2021, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Public sector", "categories": [ "Public sector" ], "system": "clinical early-warning model", "source": "https://www.statnews.com/2021/06/21/epic-sepsis-algorithm-investigation-michigan-medicine/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-036", "title": "Twitter abandoned its photo-cropping algorithm after white-face bias went viral", "summary": "Twitter's photo-preview cropper kept centering white faces over Black ones, and the company eventually scrapped the algorithm after its own review confirmed the bias.", "authority": "media-documented incident", "anchor": "Sep 2020 users showed Twitter's saliency auto-crop favored lighter faces; Twitter's audit confirmed bias and it dropped the algorithm (2021).", "jurisdiction": "Global", "country": "Global", "year": 2021, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "image saliency / auto-cropping", "source": "https://www.theguardian.com/technology/2020/sep/21/twitter-apologises-for-racist-image-cropping-algorithm", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-039", "title": "ShotSpotter alert helped jail a Chicago man for a year on a murder he didn't commit", "summary": "A man sat in jail nearly a year on a murder charge resting heavily on a gunshot-detection alert before prosecutors dropped the case.", "authority": "media-documented incident", "anchor": "AP (Aug 2021): Michael Williams spent nearly a year jailed largely on a ShotSpotter alert before charges were dismissed.", "jurisdiction": "United States", "country": "United States", "year": 2021, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "acoustic gunshot-detection", "source": "https://apnews.com/article/artificial-intelligence-algorithm-technology-police-crime-7e3345485aa668c97606d4b54f9b6220", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB2-043", "title": "Michael Oliver wrongful arrest (Detroit DataWorks Plus false match)", "summary": "Oliver was wrongfully arrested in 2019 after a DataWorks Plus facial-recognition match and sued the City of Detroit.", "authority": "U.S. District Court, E.D. Michigan", "anchor": "Civil-rights suit (filed 2020) over July 2019 wrongful larceny arrest", "jurisdiction": "United States (Michigan)", "country": "United States", "year": 2020, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.aclu.org/press-releases/aclu-statement-second-wrongful-arrest-due-face-recognition-technology", "caveat": "Suit filed; a final judgment/settlement amount for Oliver specifically not confirmed.", "cve": "" }, { "id": "LIMEN-WEB2-048", "title": "R (Bridges) v Chief Constable of South Wales Police (live facial recognition unlawful)", "summary": "The UK Court of Appeal held South Wales Police's live automated facial recognition deployment unlawful for lacking an adequate legal framework, a proper DPIA, and equality-duty compliance.", "authority": "England and Wales Court of Appeal (Civil Division)", "anchor": "[2020] EWCA Civ 1058 (Aug. 11, 2020)", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2020, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial recognition", "source": "https://www.libertyhumanrights.org.uk/issue/legal-challenge-ed-bridges-v-south-wales-police/", "caveat": "Turned on the legal framework/safeguards, not a specific wrongful identification; SWP later resumed under revised policies.", "cve": "" }, { "id": "LIMEN-WEB3-001", "title": "Netherlands SyRI welfare-fraud risk-indication system struck down", "summary": "A Dutch court ruled the SyRI welfare-fraud detection legislation breached Article 8 ECHR and ordered the system stopped.", "authority": "District Court of The Hague", "anchor": "Judgment 5 Feb 2020, ECLI:NL:RBDHA:2020:865", "jurisdiction": "Netherlands", "country": "Netherlands", "year": 2020, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "cross-database welfare/tax fraud risk profiling", "source": "https://www.loc.gov/item/global-legal-monitor/2020-03-13/netherlands-court-prohibits-governments-use-of-ai-software-to-detect-welfare-fraud/", "caveat": "First-instance ruling on legislative proportionality, not a technical audit of model bias.", "cve": "" }, { "id": "LIMEN-WEB3-009", "title": "Sweden Trelleborg automated welfare-decision source-code disclosure ruling", "summary": "After FOI litigation a Swedish administrative court held that Trelleborg's automated social-assistance decision system's source code was an official public document subject to disclosure.", "authority": "Swedish Administrative Court", "anchor": "Administrative-court FOI ruling ~2019-2020 (source code is a public document)", "jurisdiction": "Sweden", "country": "Sweden", "year": 2020, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Public sector", "categories": [ "Public sector" ], "system": "RPA for social-assistance eligibility decisions", "source": "https://algorithmwatch.org/en/trelleborg-sweden-algorithm/", "caveat": "Transparency/FOI ruling, not a finding the system itself was unlawful; exact court/date not pinned to a primary citation.", "cve": "" }, { "id": "LIMEN-WEB3-010", "title": "UK Home Office visa-application 'streaming tool' algorithm withdrawn", "summary": "Facing a judicial review alleging racial discrimination under the Equality Act, the Home Office committed in Aug 2020 to suspend and redesign its visa-streaming algorithm.", "authority": "High Court judicial review (JCWI/Foxglove); Home Office", "anchor": "Home Office agreed to discontinue the tool, 3-4 Aug 2020", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2020, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Public sector", "categories": [ "Public sector" ], "system": "nationality-based visa risk 'streaming' tool", "source": "https://www.foxglove.org.uk/2020/08/04/home-office-says-it-will-abandon-its-racist-visa-algorithm-after-we-sued-them/", "caveat": "No merits judgment; withdrawn pre-hearing, so unlawfulness was alleged not adjudicated.", "cve": "" }, { "id": "LIMEN-FAME-044", "title": "Babylon Health's AI symptom-checker accused of missing heart attacks", "summary": "A health chatbot marketed as rivaling doctors was shown failing to flag heart-attack symptoms and giving women weaker emergency advice than identical men.", "authority": "media-documented incident", "anchor": "~2018-2020 an NHS clinician documented Babylon's chatbot missing serious conditions and giving gender-skewed advice.", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": 2020, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims" ], "system": "medical symptom-checker chatbot", "source": "https://techcrunch.com/2020/02/25/first-do-no-harm/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB3-008", "title": "Poland unemployment-profiling scoring system ruled unconstitutional", "summary": "Poland's Constitutional Tribunal held the unemployment-profiling scheme unconstitutional because key parameters were set by ministerial ordinance rather than statute; the system was scrapped.", "authority": "Constitutional Tribunal", "anchor": "Judgment K 53/16, 6 June 2018; system scrapped by Dec 2019", "jurisdiction": "Poland", "country": "Poland", "year": 2019, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Public sector", "categories": [ "Public sector" ], "system": "automated jobseeker profiling/scoring", "source": "https://trybunal.gov.pl/postepowanie-i-orzeczenia/komunikaty-prasowe/komunikaty-po/art/10168-zarzadzanie-pomoca-kierowana-do-osob-bezrobotnych", "caveat": "Ruling rested on legality/separation-of-powers, not a finding of algorithmic discrimination.", "cve": "" }, { "id": "LIMEN-FAME-031", "title": "Healthcare risk algorithm (Optum) systematically under-flagged Black patients", "summary": "A health algorithm used by hospitals nationwide judged sicker Black patients to be healthier because it equated low past spending with low need.", "authority": "media-documented incident", "anchor": "Science (Oct 2019): a widely used care-management algorithm scored equally-sick Black patients as lower risk by using past spending as a proxy.", "jurisdiction": "United States", "country": "United States", "year": 2019, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Public sector", "categories": [ "Public sector" ], "system": "population-health risk-stratification", "source": "https://www.science.org/doi/10.1126/science.aax2342", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB3-024", "title": "Ark. Dep't of Human Servs. v. Ledgerwood (Arkansas Medicaid ARChoices algorithm)", "summary": "Arkansas state courts invalidated/enjoined the algorithmic ARChoices hours-allocation rule for failing rulemaking/notice requirements, after cuts averaging ~43% of attendant-care hours.", "authority": "Arkansas Supreme Court; Pulaski County Circuit Court", "anchor": "530 S.W.3d 336 (2017); rule invalidated/enjoined 2018", "jurisdiction": "United States (Arkansas)", "country": "United States", "year": 2018, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "algorithmic home-care-hours allocation tool", "source": "https://case-law.vlex.com/vid/ark-dep-t-of-894108726", "caveat": "Decided on administrative-procedure/notice grounds, not a holding that the algorithm was statistically biased.", "cve": "" }, { "id": "LIMEN-FAME-025", "title": "Amazon scraps secret AI recruiting tool that penalized women", "summary": "An experimental Amazon hiring AI trained on a decade of mostly-male resumes taught itself to favor men, and the company quietly killed it.", "authority": "media-documented incident", "anchor": "Reuters (Oct 2018): Amazon abandoned a resume-screening engine that downgraded resumes mentioning 'women's'; built 2014, scrapped ~2017.", "jurisdiction": "United States", "country": "United States", "year": 2018, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "resume-screening ML", "source": "https://www.reuters.com/article/world/insight-amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK0AG/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-029", "title": "IBM Watson for Oncology gave 'unsafe and incorrect' cancer advice", "summary": "An AI marketed as a cancer-treatment advisor produced recommendations doctors flagged as unsafe, with internal documents blaming its synthetic training cases.", "authority": "media-documented incident", "anchor": "STAT (Jul 2018): internal docs showed Watson for Oncology recommending unsafe/incorrect treatments, trained on synthetic cases; MD Anderson shelved its project (2017).", "jurisdiction": "United States", "country": "United States", "year": 2018, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Public sector", "categories": [ "Public sector" ], "system": "clinical decision-support (oncology)", "source": "https://www.statnews.com/2018/07/25/ibm-watson-recommended-unsafe-incorrect-treatments/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-035", "title": "Gender Shades study exposed commercial face-analysis bias against dark-skinned women", "summary": "A landmark audit showed commercial face software nearly perfect on light-skinned men but wrong about a third of the time on dark-skinned women.", "authority": "media-documented incident", "anchor": "2018 Buolamwini & Gebru: commercial gender classifiers erred up to 34.7% on darker-skinned women vs 0.8% on lighter-skinned men.", "jurisdiction": "Global", "country": "Global", "year": 2018, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-analysis / gender classification", "source": "https://proceedings.mlr.press/v81/buolamwini18a.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB3-012", "title": "Houston (HISD) EVAAS teacher value-added evaluation algorithm", "summary": "A US federal court held that using secret, unverifiable EVAAS value-added scores to terminate Houston teachers raised a viable procedural-due-process claim; the district then abandoned the practice in settlement.", "authority": "U.S. District Court S.D. Texas", "anchor": "Houston Fed'n of Teachers v. HISD, 251 F.Supp.3d 1168 (S.D. Tex. 2017); settled Oct 2017", "jurisdiction": "United States (Texas)", "country": "United States", "year": 2017, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "proprietary value-added teacher-scoring model (EVAAS)", "source": "https://law.justia.com/cases/federal/district-courts/texas/txsdce/4:2014cv01189/1174554/", "caveat": "Denial of summary judgment (claim survived), not a final merits finding; resolved by settlement.", "cve": "" }, { "id": "LIMEN-FAME-033", "title": "Knightscope security robot 'drowned' itself in a DC fountain", "summary": "A patrol robot drove itself down a flight of steps into a fountain after its navigation missed the uneven ground, becoming an internet meme.", "authority": "media-documented incident", "anchor": "Jul 2017 a Knightscope K5 in DC rolled down steps into a fountain; review blamed an algorithm that missed the uneven surface.", "jurisdiction": "United States", "country": "United States", "year": 2017, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous security robot", "source": "https://time.com/4862263/security-robot-fountain-knightscope-k5/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-043", "title": "Apple Face ID could not tell some Chinese users apart, triggering refunds", "summary": "Reports surfaced of Apple's Face ID unlocking phones for the wrong person among Chinese users, reviving worries about race and face recognition.", "authority": "media-documented incident", "anchor": "Dec 2017 a Chinese woman reported colleagues could unlock her iPhone X via Face ID; refunds issued; bias debate ensued.", "jurisdiction": "China", "country": "China", "year": 2017, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition authentication", "source": "https://www.theguardian.com/technology/2017/dec/22/apple-faceid-iphonex-chinese-users-unlock", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-WEB3-011", "title": "Idaho Medicaid benefit-cut algorithm (K.W. v. Armstrong)", "summary": "A US federal court held Idaho's secret budget-calculation formula for Medicaid disability assistance arbitrarily deprived participants of property without due process and ordered it disclosed with safeguards.", "authority": "U.S. District Court D. Idaho; 9th Circuit", "anchor": "9th Cir. 5 June 2015; D. Idaho due-process decision 29 Mar 2016", "jurisdiction": "United States (Idaho)", "country": "United States", "year": 2016, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector" ], "system": "automated Medicaid budget-calculation tool", "source": "https://law.justia.com/cases/federal/appellate-courts/ca9/14-35296/14-35296-2015-06-05.html", "caveat": "Tool was a statistical formula not a learned ML model; rulings turned on due-process/transparency, not bias.", "cve": "" }, { "id": "LIMEN-WEB3-023", "title": "State v. Loomis (COMPAS risk-assessment at sentencing)", "summary": "The Wisconsin Supreme Court upheld use of COMPAS at sentencing against a due-process challenge but imposed limits and required cautionary warnings about its proprietary, group-based nature.", "authority": "Supreme Court of Wisconsin", "anchor": "2016 WI 68, decided 13 July 2016", "jurisdiction": "United States (Wisconsin)", "country": "United States", "year": 2016, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "proprietary recidivism risk-assessment (COMPAS)", "source": "https://www.wicourts.gov/sc/opinion/DisplayDocument.pdf?content=pdf&seqNo=171690", "caveat": "Ruling permitted the tool rather than finding it biased; ProPublica-style bias claims were not adjudicated as fact.", "cve": "" }, { "id": "LIMEN-FAME-001", "title": "Microsoft's Tay chatbot turned racist in under 24 hours", "summary": "Microsoft's Tay chatbot was taught by Twitter users to post racist and antisemitic messages within hours of launch and was shut down the same day.", "authority": "media-documented incident", "anchor": "Microsoft launched Twitter chatbot Tay (Mar 2016); trolls taught it racist/antisemitic tweets; pulled within a day.", "jurisdiction": "United States", "country": "United States", "year": 2016, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Rights & ethics", "categories": [ "Rights & ethics" ], "system": "social-media chatbot", "source": "https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-032", "title": "Knightscope security robot knocked down and ran over a toddler", "summary": "An autonomous mall security robot bowled over a toddler and rolled across his foot despite its obstacle sensors.", "authority": "media-documented incident", "anchor": "Jul 2016 a 300-pound Knightscope K5 at Stanford Shopping Center knocked down a 16-month-old and rolled over his foot.", "jurisdiction": "United States", "country": "United States", "year": 2016, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Agentic / control failure", "categories": [ "Agentic / control failure" ], "system": "autonomous security robot", "source": "https://money.cnn.com/2016/07/14/technology/robot-stanford-mall/index.html", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-040", "title": "PredPol predictive policing accused of creating racist feedback loops", "summary": "A crime-prediction tool learned from biased arrest data and kept steering police to the same neighborhoods, predicting policing more than crime.", "authority": "media-documented incident", "anchor": "Reporting/research (2016+) found PredPol predicted future policing rather than crime, reinforcing patrols in over-policed minority areas.", "jurisdiction": "United States", "country": "United States", "year": 2016, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "predictive-policing hotspot model", "source": "https://www.technologyreview.com/2020/07/17/1005396/predictive-policing-algorithms-racist-dismantled-machine-learning-bias-criminal-justice/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-041", "title": "Chicago's 'heat list' flagged thousands yet failed to curb violence", "summary": "A secret police algorithm ranked over a thousand mostly-Black Chicagoans by risk, but a study found it cut no violence while raising their odds of arrest.", "authority": "media-documented incident", "anchor": "2016 RAND study: Chicago PD's Strategic Subject List (~1,000+ people, ~77% Black) didn't reduce victimization but raised arrest odds.", "jurisdiction": "United States", "country": "United States", "year": 2016, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "individual risk-scoring 'heat list'", "source": "https://www.chicagomag.com/city-life/August-2016/Chicago-Police-Data/", "caveat": "Media-documented incident, not a regulator/court finding. | VERIFIED: The RAND-evaluated Strategic Subjects List version had 426 people (77% Black, 95.8% male); the larger ~1,000-1,400 figures come from later/other accounts, not the RAND study.", "cve": "" }, { "id": "LIMEN-FAME-045", "title": "COMPAS recidivism scores shown biased against Black defendants", "summary": "An investigation found a courtroom risk-scoring tool falsely tagged Black defendants as future criminals at roughly twice the rate of white ones.", "authority": "media-documented incident", "anchor": "ProPublica (May 2016): COMPAS wrongly labeled Black defendants high-risk at ~2x the rate of white defendants who didn't reoffend; vendor disputed.", "jurisdiction": "United States", "country": "United States", "year": 2016, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "criminal-risk / recidivism scoring", "source": "https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing", "caveat": "Media-documented incident (journalistic analysis), not a court ruling.", "cve": "" }, { "id": "LIMEN-FAME-027", "title": "Google Photos labelled two Black users as 'gorillas'", "summary": "Google Photos' image recognition tagged Black people as gorillas, and the company's lasting fix was to delete the label rather than fix the model.", "authority": "media-documented incident", "anchor": "Jun/Jul 2015 a photo of two Black people was auto-tagged 'gorillas'; Google 'fixed' it by blocking the label.", "jurisdiction": "Global", "country": "Global", "year": 2015, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "image classification / auto-tagging", "source": "https://www.bbc.com/news/technology-33347866", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-038", "title": "Knight Capital's runaway trading algorithm lost $440M in 45 minutes", "summary": "A bad software rollout let a trading firm's automated system flood the market with errant orders, costing it about $440 million in under an hour.", "authority": "media-documented incident", "anchor": "1 Aug 2012 a botched code deployment fired millions of erroneous orders across 154 stocks, a ~$440M loss in 45 minutes.", "jurisdiction": "United States", "country": "United States", "year": 2012, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "automated equities trading", "source": "https://dealbook.nytimes.com/2012/08/02/knight-capital-says-trading-mishap-cost-it-440-million/", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-FAME-037", "title": "The 2010 'Flash Crash': algorithms erased ~$1 trillion in minutes", "summary": "Automated trading turned a single large sell order into a few-minute market plunge that briefly wiped out roughly a trillion dollars before bouncing back.", "authority": "media-documented incident", "anchor": "6 May 2010 the Dow fell ~1,000 points in minutes; SEC/CFTC tied the cascade to a large automated sell order and HFT.", "jurisdiction": "United States", "country": "United States", "year": 2010, "evidence": "Notable incident (media)", "tier": "notable_incident_media_documented", "category": "Finance & insurance", "categories": [ "Finance & insurance" ], "system": "algorithmic / high-frequency trading", "source": "https://www.sec.gov/news/studies/2010/marketevents-report.pdf", "caveat": "Media-documented incident, not a regulator/court finding.", "cve": "" }, { "id": "LIMEN-000001", "title": "Auto-GPT sandbox escape and host code execution risk via unsanitized file basename", "summary": "Supports a bounded technical-security observation and a hostile-reviewer-safe crosswalk row showing technique support without a canonical AVID incident ID.", "authority": "authoritative", "anchor": "LIMEN-S-006", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2023/37xxx/CVE-2023-37274.json", "caveat": "keep AVID linkage negative/indirect unless a canonical AVID report appears; do not narrate AML.CS0052 as a CVE-specific match", "cve": "CVE-2023-37274" }, { "id": "LIMEN-000002", "title": "Langflow custom_component endpoint allowed remote code execution; GitHub reviewed advisory now exposes first patched version 1.0.15", "summary": "The row now supports reviewer-safe exact-AVID interoperability plus formal advisory and exact-fix visibility, promoting Langflow from a limitations-only notice-gap exemplar into an appendix-grade remediation-comparison row without claiming a direct ATLAS case or broader prevalence.", "authority": "authoritative", "anchor": "LIMEN-S-089", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent", "source": "https://api.github.com/advisories/GHSA-qg33-x2c5-6p44", "caveat": "no direct ATLAS same-case match was verified in the local package; keep the GitHub-advisory `< 1.0.15` framing and the CVE/OSV `through 0.6.19` framing source-qualified rather than forcing one unified affected-range statement", "cve": "CVE-2024-37014" }, { "id": "LIMEN-000003", "title": "Claude Code default allowlist enabled unauthorized file read and network exfiltration before forced update", "summary": "Supports a bounded approval-model failure argument plus a reviewer-safe crosswalk row where technique taxonomy is stronger than exact cross-framework incident IDs.", "authority": "authoritative", "anchor": "LIMEN-S-008", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, coding assistant", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2025/55xxx/CVE-2025-55284.json", "caveat": "keep framework linkage at technique/taxonomy level until a direct AVID report or ATLAS case-study exists", "cve": "CVE-2025-55284" }, { "id": "LIMEN-000004", "title": "CrewAI JSON loader allowed arbitrary local file read on the server", "summary": "Supports a bounded file-exposure/tool-misuse row with exact fixed-version visibility and an exact vendor fix-path statement, while still keeping advisory-depth claims below formal security-bulletin grade.", "authority": "authoritative", "anchor": "LIMEN-S-009", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, multi agent", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/2xxx/CVE-2026-2285.json", "caveat": "keep framework linkage at technique/taxonomy level only; treat the CVE-linked PR plus release/tag proof as exact-fix visibility below formal advisory grade rather than as a full vendor advisory", "cve": "CVE-2026-2285" }, { "id": "LIMEN-000005", "title": "CrewAI RAG search tools enabled server-side request forgery and internal/cloud content acquisition", "summary": "Supports a bounded retrieval/network-surface risk row with exact fixed-version visibility and an exact vendor fix-path statement, while still keeping advisory-depth claims below formal security-bulletin grade.", "authority": "authoritative", "anchor": "LIMEN-S-010", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, multi agent", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/2xxx/CVE-2026-2286.json", "caveat": "keep framework linkage at technique/taxonomy level only; treat the CVE-linked PR plus release/tag proof as exact-fix visibility below formal advisory grade rather than as a full vendor advisory", "cve": "CVE-2026-2286" }, { "id": "LIMEN-000006", "title": "LangChain LLMMathChain allowed prompt injection into Python exec, creating arbitrary code execution risk", "summary": "Adds the package's first independent-lab public-security exemplar: primary CVE record plus third-party Snyk advisory, which strengthens source-independence discussion without implying prevalence or complete remediation certainty.", "authority": "authoritative", "anchor": "LIMEN-S-032", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, framework", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2023/29xxx/CVE-2023-29374.json", "caveat": "no direct vendor-authored advisory was verified in the bounded pass; keep fixed-version language source-qualified because the public CVE scope text and Snyk upgrade guidance are not phrased identically; keep ATLAS linkage at technique level only", "cve": "CVE-2023-29374" }, { "id": "LIMEN-000007", "title": "LlamaIndex safe_eval command injection enabled LLM-generated code to execute OS commands on the host", "summary": "Adds a second independent-lab public-security exemplar: primary CVE record plus public Snyk advisory for a framework-level unsafe-code-execution path, which strengthens source-independence discussion without implying clean vendor remediation visibility or prevalence.", "authority": "authoritative", "anchor": "LIMEN-S-036", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, framework", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2024/3xxx/CVE-2024-3271.json", "caveat": "no direct vendor-authored advisory was verified in the bounded pass; keep fixed-version language source-qualified because the CVE affected-version notation and Snyk upgrade guidance are not phrased identically; keep ATLAS linkage at technique level only", "cve": "CVE-2024-3271" }, { "id": "LIMEN-000008", "title": "LangChain broad load() allowlists enabled attacker-controlled object revival, chat-history poisoning, and possible credential or request side effects", "summary": "Adds a vendor-authored public-security exemplar for memory and state contamination rather than another code-execution row, which broadens the package's reviewer-safe security taxonomy without implying prevalence or incident realization.", "authority": "authoritative", "anchor": "LIMEN-S-041", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, framework", "source": "https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/44xxx/CVE-2026-44843.json", "caveat": "keep ATLAS linkage cautious because the local package has no exact case-study match; no direct AVID incident ID was verified in the bounded pass; do not treat the advisory's impact examples as realized incident evidence", "cve": "CVE-2026-44843" }, { "id": "LIMEN-000009", "title": "PraisonAI Platform workspace member could add arbitrary user as owner, enabling agent-workspace identity abuse", "summary": "Adds the package's first public agent-identity / authorization-boundary exemplar, reducing a previously explicit zero state in the security balance matrix without implying prevalence or realized compromise.", "authority": "authoritative", "anchor": "LIMEN-S-045", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, multi agent, platform", "source": "https://api.github.com/repos/MervinPraison/PraisonAI/security-advisories/GHSA-8g2p-pqm3-fcfh", "caveat": "No direct local AVID report or exact ATLAS case-study link was verified in the bounded pass; public CVE raw/NVD visibility was not independently resolved from this lane, so keep the row grounded in the upstream GitHub advisory and patched-version metadata only.", "cve": "CVE-2026-47413" }, { "id": "LIMEN-000012", "title": "Playwright MCP origin validation flaw allowed unauthorized local MCP requests via DNS rebinding before 0.0.40", "summary": "Adds the package's first seeded MCP / plugin-style trust-boundary exemplar, removing the previous supply-chain-or-plugin zero state while keeping the row appendix-routed and release-note-qualified rather than advisory-grade.", "authority": "authoritative", "anchor": "LIMEN-S-051", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, developer tool, mcp server", "source": "https://nvd.nist.gov/vuln/detail/CVE-2025-9611", "caveat": "No direct local AVID report or exact ATLAS case-study link was verified in the bounded pass; keep the row grounded in the public vulnerability record plus vendor release note, and do not narrate the case as realized exploitation in the wild or as broad MCP ecosystem prevalence.", "cve": "CVE-2025-9611" }, { "id": "LIMEN-000016", "title": "ToolHive MCP deployment utility stored secrets in run config files before 0.0.33, exposing local secret material", "summary": "Adds a second seeded MCP / supply-chain trust-boundary exemplar with exact AVID support and vendor-advisory-grade remediation visibility, letting the package discuss MCP deployment-trust risks without relying on Playwright MCP alone.", "authority": "authoritative", "anchor": "LIMEN-S-071", "jurisdiction": "Global", "country": "Global", "year": null, "evidence": "Security disclosure (CVE)", "tier": "T3_authoritative_source", "category": "Security & agentic", "categories": [ "Security & agentic", "Agentic / control failure" ], "system": "agent, developer tool, mcp server", "source": "https://api.github.com/repos/stacklok/toolhive/security-advisories/GHSA-xj5p-w2v5-fjm6", "caveat": "No exact ATLAS case-study match was verified in the local package; keep the home-folder-access precondition explicit and do not generalize one MCP deployment-tool secret-exposure row into broad MCP-ecosystem prevalence.", "cve": "CVE-2025-47274" }, { "id": "LIMEN-AUTH-000001", "title": "Italian DPA orders urgent provisional limitation on ChatGPT personal-data processing in Italy", "summary": "Authoritative regulator order concerning ChatGPT data-processing practices, framed by the Italian DPA as an urgent provisional limitation pending further inquiry.", "authority": "authoritative", "anchor": "SRC-AUTH-0001", "jurisdiction": "Italy", "country": "Italy", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural", "Rights & ethics" ], "system": "chatbot", "source": "", "caveat": "Use the regulator's own language: urgent provisional limitation, not a generalized finding of illegality beyond the order text.; Italian original should be preserved for citation and legal nuance review.", "cve": "" }, { "id": "LIMEN-AUTH-000003", "title": "UK ICO orders Serco Leisure and associated trusts to stop biometric attendance monitoring", "summary": "Authoritative UK regulator enforcement package with direct enforcement-notice text concerning facial recognition and fingerprint scanning used for employee attendance monitoring.", "authority": "authoritative", "anchor": "SRC-AUTH-0003", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Hiring & education", "categories": [ "Hiring & education", "Surveillance & biometrics", "Public sector" ], "system": "biometric", "source": "", "caveat": "Keep to the ICO's proportionality/fairness framing and enforcement-notice description; do not generalize beyond the notice summary page.; The enforcement-action page adds bounded scale/detail (more than 2,000 employees; 38 facilities) and record type (enforcement notices) without changing the need to avoid broader legality or sector-prevalence claims.; Direct notice text is now locally captured, so narrow claims can cite the ICO notice itself rather than only the newsroom and enforcement-record summaries.", "cve": "" }, { "id": "LIMEN-AUTH-000004", "title": "UK ICO reprimands Essex school over facial-recognition canteen payments", "summary": "Authoritative UK regulator reprimand package with direct reprimand text concerning school use of facial recognition technology for student canteen payments.", "authority": "authoritative", "anchor": "SRC-AUTH-0004", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics", "Public sector", "Hiring & education", "Rights & ethics" ], "system": "biometric", "source": "", "caveat": "The source frames this as a reprimand and data-protection failure; avoid inferring broader school-sector prevalence from one case.; Student/minor context warrants metadata-first treatment.; The enforcement-action page provides the narrower legal framing of a reprimand in respect of Article 35(1) tied to failure to complete a DPIA before introduction.; The linked newsroom page is dated 2024-07-23 while the enforcement-action record is dated 2024-07-22; keep both dates explicit rather than collapsing them.; Direct reprimand text is now locally captured, which supports narrow citation of the ICO reprimand itself while still avoiding broader school-sector inference.", "cve": "" }, { "id": "LIMEN-AUTH-000005", "title": "DOJ charges Nate founder over alleged AI investment-fraud misstatements and hidden human transaction processing", "summary": "Authoritative DOJ charging-stage package with a linked indictment PDF concerning alleged false statements about nate's proprietary AI technology and operational capabilities.", "authority": "authoritative_charging_stage", "anchor": "SRC-AUTH-0010", "jurisdiction": "United States", "country": "United States", "year": null, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Surveillance & biometrics" ], "system": "other", "source": "", "caveat": "Use this row as charging-stage authority only; do not convert the indictment allegations into adjudicated fact or liability.; Carry the DOJ presumption-of-innocence sentence with any manuscript or dashboard reuse.; The linked indictment PDF improves court-facing document presence, but exact caption/case-number extraction still remains a next-step normalization task. | VERIFIED: DOJ/SEC actions against Nate founder Albert Saniger are an April 2025 indictment and unproven civil charges (charging stage), not an adjudicated finding.", "cve": "" }, { "id": "LIMEN-AUTH-000006", "title": "SEC settlement with Delphia over false and misleading statements about AI and machine learning use", "summary": "The SEC found that Delphia made false and misleading statements in filings, advertisements, and social media about using artificial intelligence and machine learning with client data in its investment process.", "authority": "authoritative", "anchor": "SRC-AUTH-0012", "jurisdiction": "United States", "country": "United States", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Finance & insurance" ], "system": "predictive model;recommendation", "source": "", "caveat": "Use as an SEC enforcement finding and settlement record, not as a judicial determination of broader fraud facts.; Good paper-grade example of AI capability exaggeration in regulated financial advice marketing.", "cve": "" }, { "id": "LIMEN-AUTH-000007", "title": "SEC settlement with Global Predictions over false and misleading AI financial-adviser claims", "summary": "The SEC found that Global Predictions made false and misleading claims about its use of AI, including describing itself as the first regulated AI financial advisor and misrepresenting certain service features.", "authority": "authoritative", "anchor": "SRC-AUTH-0014", "jurisdiction": "United States", "country": "United States", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Finance & insurance" ], "system": "recommendation;chatbot", "source": "", "caveat": "Use as an SEC enforcement finding and settlement record, not as proof that all AI-enabled advisory products are similarly misleading.; Useful comparator case for paper tables on AI washing in regulated sectors.", "cve": "" }, { "id": "LIMEN-AUTH-000010", "title": "FTC final order and complaint package on DoNotPay AI-lawyer and robot-lawyer claims", "summary": "Official FTC complaint-plus-final-order package on DoNotPay's AI-lawyer / robot-lawyer capability claims, now promoted into the shared authoritative core as a bounded non-finance official AI-claim exemplar rather than a judicial unauthorized-practice finding.", "authority": "authoritative", "anchor": "SRC-AUTH-0020", "jurisdiction": "United States", "country": "United States", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Legal & procedural" ], "system": "chatbot;document generation;legal service automation", "source": "", "caveat": "Use FTC complaint/final-order posture only; do not restate the row as a judicial unauthorized-practice finding, adjudicated product-truth finding, or general legal-services automation verdict.; The row is strongest as a bounded non-finance official AI-claim exemplar that widens authority diversification beyond settled SEC finance matters.; Keep complaint allegations and final-order prohibitions distinct when reusing this package in manuscript, dashboard, or thesis prose.", "cve": "" }, { "id": "LIMEN-AUTH-000011", "title": "FTC final order and complaint package on Workado / Content at Scale AI detector accuracy claims", "summary": "Official FTC complaint-plus-final-order package on Workado / Content at Scale AI detector-accuracy advertising, now promoted into the shared authoritative core as a bounded detector-governance and false-capability row.", "authority": "authoritative", "anchor": "SRC-AUTH-0022", "jurisdiction": "United States", "country": "United States", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "AI-washing & false claims", "categories": [ "AI-washing & false claims", "Hiring & education" ], "system": "ai detector;text classification", "source": "", "caveat": "Use FTC complaint/final-order posture only; do not convert this row into a general scientific-validity conclusion about all AI detectors or a prevalence claim about detector fraud.; The row is strongest as a bounded detector-governance / false-capability exemplar that can bridge AI-misstatement evidence with research-integrity and educational-process caution without overclaiming causality.; Keep the complaint's performance allegations distinct from the final order's substantiation and recordkeeping requirements.", "cve": "" }, { "id": "LIMEN-AUTH-000012", "title": "Datatilsynet development-and-drift opinions on AI-supported screening of disability-allowance documentation in Danish student-aid administration", "summary": "Official Danish regulator development-and-drift page pair plus a case-specific legal-basis chain and official UFS taskforce report, with live readable Retsinformation legal text confirming both the row-specific disability-allowance screening context and a broader ministerial authorization for AI or similar digital solutions in SU applications and related case handling, preserved as a rare-language public-sector governance anchor rather than a harm or deployment finding.", "authority": "authoritative_candidate", "anchor": "SRC-AUTH-0024", "jurisdiction": "Denmark", "country": "Denmark", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Public sector", "categories": [ "Public sector", "Hiring & education" ], "system": "decision support;document screening;public sector benefits administration", "source": "", "caveat": "Preserve Danish original and treat any English gloss as a translation-aided paraphrase rather than a final legal interpretation.; Use the page as a governance/legal-basis anchor only; it is not, by itself, evidence of harm, illegality, deployment failure, or compliance certification.; A 2025 same-row Datatilsynet drift page now provides exact-row follow-through beyond the 2024 development page, but it still supports governance/legal-basis claims rather than procurement proof, impact evidence, or general legality conclusions.; The Danish row now has a specific statutory-plus-regulatory companion chain (Section 39 a plus BEK nr 813), but this remains governance/legal-basis support rather than procurement, evaluation, impact, harm, or compliance proof.", "cve": "" }, { "id": "LIMEN-AUTH-000015", "title": "CNIL warns sports club over proposed facial-recognition checks for commercial stadium bans", "summary": "Official French CNIL warning-page candidate on proposed facial-recognition use by a sports club for commercial stadium-ban enforcement, preserved as a bounded non-U.S./non-UK biometric-governance and legality-posture row.", "authority": "authoritative_candidate", "anchor": "SRC-AUTH-0031", "jurisdiction": "France", "country": "France", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "biometric;facial recognition", "source": "", "caveat": "Use the CNIL warning-page posture only; do not infer that the contemplated system was actually deployed or that a pecuniary sanction was imposed.; Preserve the French original for legality wording and do not widen the page into a general facial-recognition prohibition beyond the stated sports-club setting and current legal-framework analysis.; The row improves non-U.S./non-UK official authority breadth, but it remains page-summary depth unless a direct warning decision or companion document is later captured.", "cve": "" }, { "id": "LIMEN-PROC-003", "title": "SAGE / ChatGPT fabricated-reference correction (research integrity)", "summary": "LIMEN can show an intermediate research-integrity class where article identity is verified, correction-status metadata is visible, and a named editor is publicly quoted, but no direct journal notice text is captured.", "authority": "publisher_or_regulator_notice", "anchor": "", "jurisdiction": "Multijurisdictional scholarly publishing", "country": "Multijurisdictional scholarly publishing", "year": null, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "", "source": "", "caveat": "Do not describe the editorial action as final or infer correction wording unless a direct journal or publisher record is captured. | VERIFIED: SAGE journal Environment and Planning E was finalizing 'corrective actions' (no correction issued as of May 2024) for ChatGPT-mangled citations that the editor said were NOT fabricated.", "cve": "" }, { "id": "LIMEN-PROC-004", "title": "NEJM AI-imagery publisher notice (research integrity)", "summary": "LIMEN can now show a non-Springer direct-notice class where archived public NEJM pages expose the authors' AI-assisted image-manipulation explanation and the journal's disclosure-policy framing.", "authority": "publisher_or_regulator_notice", "anchor": "", "jurisdiction": "Multijurisdictional scholarly publishing", "country": "Multijurisdictional scholarly publishing", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "", "source": "", "caveat": "Use the archived notice text narrowly and do not generalize into broader misconduct, clinical significance, or prevalence claims. | VERIFIED: NEJM fully retracted 'Bronchial Casts from Inhalation of Forest-Fire Smoke' (N Engl J Med 2026;394:1634) ~April 2026 over an undisclosed AI-manipulated figure.", "cve": "" }, { "id": "LIMEN-PROC-005", "title": "Springer book fabricated-citations notice (research integrity)", "summary": "LIMEN now has a second direct notice-text exemplar in shard 008, broadening direct publisher-originating evidence from a journal article to a book surface.", "authority": "publisher_or_regulator_notice", "anchor": "", "jurisdiction": "Multijurisdictional scholarly publishing", "country": "Multijurisdictional scholarly publishing", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "", "source": "", "caveat": "Use only the notice-bounded rationale about unverifiable references; do not infer generative-AI causation, intent, or broader publisher practice from this one book notice. | VERIFIED: Springer Nature retracted 'Mastering Machine Learning: From Basics to Advanced' (Govindakumar Madhavan, Apr 2025) in Aug 2025 after 25 of 46 references were unverifiable.", "cve": "" }, { "id": "LIMEN-PROC-006", "title": "Springer ICM fabricated-references notice (research integrity)", "summary": "LIMEN has one direct notice-text exemplar that supports a narrow publisher-originating citation-contamination claim.", "authority": "publisher_or_regulator_notice", "anchor": "", "jurisdiction": "Multijurisdictional scholarly publishing", "country": "Multijurisdictional scholarly publishing", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Legal & procedural", "categories": [ "Legal & procedural" ], "system": "", "source": "", "caveat": "Use only the notice-bounded rationale on the page and do not generalize across publishers, venues, or author intent. | VERIFIED: Intensive Care Medicine (Springer Nature) retracted a Dec 2024 letter on Nov 29 after 10 of its 15 AI-generated references proved nonexistent.", "cve": "" }, { "id": "LIMEN-WEB4-012", "title": "Spain fines a sports-centre operator for mandatory facial-recognition access", "summary": "AEPD fined a sports-centre operator EUR 96,000 for mandatory facial-recognition access without consent.", "authority": "AEPD", "anchor": "EXP202213792, EUR 96,000", "jurisdiction": "Spain", "country": "Spain", "year": null, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Surveillance & biometrics", "categories": [ "Surveillance & biometrics" ], "system": "facial-recognition access control", "source": "https://gdprhub.eu/index.php?title=AEPD_(Spain)_-_EXP202213792", "caveat": "GDPRhub page not directly fetched. | VERIFIED: The EUR 96,000 gym fine belongs to the SIDECU/Supera chain (EUR 160,000 reduced to 96,000, decision ~Aug 2025); case ID EXP202213792 actually identifies the Burgos CF stadium case (duplicate of LIMEN-WEB4-010), so the anchor pairs the wrong file number with this fine.", "cve": "" }, { "id": "LIMEN-WEB4-021", "title": "Turkey bars and fines biometric facial-recognition attendance by employers", "summary": "The Turkish KVKK ruled biometric attendance unlawful even with consent and imposed administrative fines.", "authority": "KVKK", "anchor": "TRY 500,000 (facial) + TRY 225,000 (palm-reading)", "jurisdiction": "Turkey", "country": "Turkey", "year": null, "evidence": "Contested / interim", "tier": "T2_contested_or_interim", "category": "Hiring & education", "categories": [ "Hiring & education" ], "system": "facial/palm-vein biometric attendance", "source": "https://www.biometricupdate.com/202606/turkiye-says-employers-cannot-rely-on-biometrics-for-attendance-monitoring", "caveat": "Decision numbers/dates not isolated.", "cve": "" }, { "id": "LIMEN-WEB4-069", "title": "R v Karl Marshall — jailed for AI deepfake intimate images of women and children", "summary": "A UK man received a custodial sentence for creating hundreds of AI deepfake intimate images.", "authority": "Crown Court; CPS", "anchor": "2y2m; 266 AI deepfake images", "jurisdiction": "United Kingdom", "country": "United Kingdom", "year": null, "evidence": "Regulator / court record", "tier": "T3_authoritative_source", "category": "Deepfakes & synthetic media", "categories": [ "Deepfakes & synthetic media" ], "system": "deepfake", "source": "https://feeds.bbci.co.uk/news/articles/cewgxd5yewjo", "caveat": "BBC URL failed direct fetch; confirm charges against official record.", "cve": "" } ] }