{"meta":{"exportedAt":"2026-10-27T06:03:27.720Z","formatVersion":2,"selection":{"q":"pay","system":"","harm":"","context":"","country":"","role":"","relation":"contextual","evidence":"","year":"","response":"","severity":"","verification":"","view":"incidents","sort":"added"},"totalIncidents":17,"coverage":{"cases":17,"countries":8,"languages":7,"unknownLocation":1,"locationPending":0,"unknownLanguage":0,"unknownDate":0,"lawsuits":3,"regulatory":3,"minors":1,"coreRelations":5,"contextualRelations":17,"mixedRelations":5,"unknownRelations":0,"relationPending":0,"relationUnknown":0},"countingNote":"Distinct public cases in this selection. People counts apply within individual cases only; cross-case person overlap has not been resolved. No population incidence estimate.","affectedCountNote":"Interpret person counts with affectedCountStatus and the reported effects. Unquantified zeros are placeholders, not a measured zero.","source":"AI incidents","publisher":"NOPE","url":"https://nope.net/incidents","license":"CC BY 4.0"},"incidents":[{"id":"2025-ramat-gan-supreme-court-finds-city-denied-special-education-pupils-school-transport-with-ai-invented-circular-and-rulings","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'נדחתה בקשתו של האב .במענה מיום  25.5.2025נכתב ,כי בהתאם להוראות \"חוזר מנכ\"ל'","relation":"supports","source_id":"s1"},{"locator":"'בתגובה, במאי 2025 עובדי העירייה שלחו מכתב רשמי ומפורט שבו דחו את בקשתו.'","relation":"supports","source_id":"s2"}],"assertion":"On 25 May 2025 the municipality refused transport for the pupil from his father's home, citing an Education Ministry director-general circular and quotations from recent rulings.","causal_attribution":"Facts recited by the Supreme Court from the documents before it."},{"id":"c2","status":"documented","evidence":[{"locator":"'המנכ\"ל אליו מפנה עיריית רמת גן אינו מוכר לנו'","relation":"supports","source_id":"s1"},{"locator":"'גם במשרד החינוך אמרו כי החוזר שהעירייה הפנתה אליו לא מוכר להם'","relation":"supports","source_id":"s2"}],"assertion":"The Education Ministry told the father's lawyer that it did not know the circular the municipality cited.","causal_attribution":"The ministry's reply as quoted in the judgment."},{"id":"c3","status":"documented","evidence":[{"locator":"'עיון במסמכים שצורפו לערעור ,לא מותיר מקום לספק באשר לשימוש שעשתה'; 'כפי שהוצגה למערער – התבססה בבירור על שימוש לא מבוקר בבינה מלאכותית'","relation":"supports","source_id":"s1"},{"locator":"'עיריית רמת גן תשלם הוצאות בסך 30,000 שקל בשל \"שימוש מופקר\"'","relation":"supports","source_id":"s3"}],"assertion":"The Supreme Court found that the documents left no doubt that the municipality had used AI, and that its decision, at least as presented to the pupil, clearly rested on uncontrolled use of artificial intelligence.","causal_attribution":"The court's own finding on the AI use; the municipality did not deny it but argued the errors were technical."},{"id":"c4","status":"documented","evidence":[{"locator":"'בתשלום הוצאות על סך של  – ₪ 30,000'","relation":"supports","source_id":"s1"},{"locator":"'ביולי העירייה הסכימה לפנים משורת הדין להחזיר לילד את הזכאות להסעות'; 'בית המשפט העליון חייב את עיריית רמת גן לשלם לאב 30,000 ש\"ח הוצאות משפט'","relation":"supports","source_id":"s2"}],"assertion":"The municipality agreed in July 2025 to restore the transport, and the Supreme Court ordered it to pay 30,000 shekels in costs.","causal_attribution":"Outcome as recorded in the judgment and reported."}],"effects":[{"label":"special-education pupil refused school transport on the basis of AI-invented authorities; family forced to litigate","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://z.calcalist.co.il/assets/pickerul/f8edf1ed-a81f-40ca-9d68-80fe1e65d86e.pdf","kind":"court_record","access":"read","language":"he","translation_note":"Read in Hebrew by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Supreme Court judgment PDF hosted by Calcalist; the live file returned HTTP 403, so a Web Archive capture of the same URL was read in full (text extracted from the PDF).","independence_group":"israel-supreme-court-judgment"},{"id":"s2","url":"https://www.davar1.co.il/665441/","kind":"news_report","access":"read","language":"he","translation_note":"Read in Hebrew by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Davar report of 24 March 2026 with comment from the father's lawyer; the live page returned HTTP 403, so a Web Archive capture of the same URL was read in full.","independence_group":"davar-report"},{"id":"s3","url":"https://www.bizportal.co.il/takdin/news/article/20029328","kind":"news_report","access":"read","language":"he","translation_note":"Read in Hebrew by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Bizportal (Takdin) report of 22 March 2026 quoting the judgment; HTML read, HTTP 200.","independence_group":"bizportal-takdin-report"}],"version":1,"ai_roles":["institutional_use"],"contexts":["education","public_services","justice"],"unknowns":["Which AI system the municipality used and who in the municipality used it.","How long the pupil was without transport from his father's home between the refusal and July 2025.","Any disciplinary outcome inside the municipality."],"geography":{"basis":"The decision was taken by the Ramat Gan municipality and litigated in the Tel Aviv District Court and the Israeli Supreme Court (judgment and Davar). The family's residence is not used as the event location.","court_countries":["IL"],"event_countries":["IL"],"affected_person_countries":[]},"publication":{"basis":"Published as an AI-informed administrative decision about a person: the Israeli Supreme Court found that Ramat Gan municipality's refusal of a special-education pupil's school transport rested on uncontrolled AI use with an invented circular and rulings, and ordered costs against the municipality.","reviewed_on":"2026-10-27"},"ai_involvement":{"basis":"The Supreme Court found that the documents left no doubt that the municipality had used AI in its initial decision and in the proceedings, and that the refusal, at least as presented to the pupil, clearly rested on uncontrolled AI use: the circular it cited was unknown to the Education Ministry and the quoted rulings did not contain the passages. The AI output thus supplied the reasons for the municipality's decision about the pupil. The specific AI system is not named.","status":"supported"},"person_relations":["informed_decision_about"]},"name":"Israel: Supreme Court finds Ramat Gan refused a special-education pupil school transport in a letter resting on uncontrolled AI use and invented rulings","summary":"In May 2025 the Ramat Gan municipality in Israel refused a request that a special-education pupil be given school transport from his father's home, citing an Education Ministry director-general circular and quotations from recent rulings. The father's lawyer could not find the circular, and the Education Ministry said it did not know it. After the father petitioned the Tel Aviv District Court, the municipality agreed in July 2025 to restore the transport but again filed invented quotations, and the district court refused costs. On appeal, the Supreme Court (Deputy President Noam Sohlberg, ruling published 22 March 2026) held that the documents left no doubt that the municipality had used artificial intelligence and that its decision, at least as presented to the pupil, clearly rested on uncontrolled AI use. It ordered the municipality to pay 30,000 shekels in costs.","incidentDate":"2025-05-25","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2026-03-22","aiSystem":"Unidentified generative AI system whose uncontrolled use, according to the Israeli Supreme Court, produced the non-existent circular and invented case-law quotations in the Ramat Gan municipality's refusal letter and court filings","aiProduct":"Unidentified AI tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["educational_harm","legal_harm","psychological_distress"],"harmOutcomeSummary":"The pupil was refused school transport from his father's home on the basis of invented authorities, and the family had to petition the courts; the father's lawyer told Davar the father felt anger and frustration. Transport was restored in July 2025.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":0,"otherPeopleHarmedMin":2,"affectedCountStatus":"exact","affectedCountEvidence":"The pupil (the appellant) and his father, who brought the request and the litigation; the ruling and Davar describe both. The municipality's staff used the AI system, so neither is counted as a user.","victimAgeRange":"minor","jurisdiction":"IL","platformType":"other","outcomeStatus":"resolved","primarySourceUrl":"https://z.calcalist.co.il/assets/pickerul/f8edf1ed-a81f-40ca-9d68-80fe1e65d86e.pdf","primarySourceLabel":"Israel Supreme Court, appeal judgment of Deputy President N. Sohlberg (Ramat Gan school transport), March 2026","firstPublishedAt":"2026-10-27T03:19:07.496358+00:00","updatedAt":"2026-10-27T03:19:07.496358+00:00","scopeVersion":"facts-v3","tags":["public-services","education","special-education","hallucinated-authorities","supreme-court","israel"]},{"id":"2021-amazon-delivery-drivers-say-netradyne-ai-van-cameras-flagged-driving-events-they-did-not-commit-lowering-bonus-scores","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'In February, Amazon announced that it would install cameras made by the AI-tech startup Netradyne in its Amazon-branded delivery vans'; 'Each time the camera registers an event, footage is uploaded into a system, recorded, and affects a score drivers receive at the end of the week for safe driving.'","relation":"supports","source_id":"s1"}],"assertion":"Amazon installed AI-powered Netradyne cameras in its delivery vans during 2021; the cameras detect driving events, upload the footage and feed a weekly safe-driving score.","causal_attribution":"Describes the system and its role in the score."},{"id":"c2","status":"reported","evidence":[{"locator":"'the camera in his van started to incorrectly penalize him whenever cars cut him off'; 'the Netradyne camera in his van has also detected “events” that didn’t actually happen, but that impacted his performance score at Amazon'","relation":"supports","source_id":"s1"}],"assertion":"Drivers said the cameras registered events they did not commit or that were not unsafe: a driver in Los Angeles said it incorrectly penalised him when cars cut him off, and a former driver in Texas said it detected events that did not happen and that they affected his performance score.","causal_attribution":"The drivers' accounts attribute the penalties to the camera's detections; Amazon and Netradyne did not address these individual events."},{"id":"c3","status":"reported","evidence":[{"locator":"'A few times, we’ve been in the country on a dirt road, where there’s no stop sign, but the camera flags a stop sign.'; 'If I look into my mirrors to make sure I am safe to change lanes, it dings me for distraction because my face is turned to look into my mirror.'","relation":"supports","source_id":"s1"}],"assertion":"Other drivers described false stop-sign detections, including a stop sign flagged on a dirt road with no sign, and being flagged for distraction when checking mirrors.","causal_attribution":"Drivers' accounts of the cameras' detection errors."},{"id":"c4","status":"reported","evidence":[{"locator":"'For many Amazon drivers, these performance scores determine whether they receive weekly bonuses, prizes, and extra pay.'; 'Before I would be able to win prizes and stuff, as soon as cameras came along, it went downhill,'; 'which receive bonuses by earning “fantastic” scores on a weekly scorecard'","relation":"supports","source_id":"s1"}],"assertion":"According to drivers and an Amazon scorecard document obtained by Motherboard, the weekly scores determine bonuses, prizes and extra pay, and the former Texas driver said winning prizes went downhill once the cameras came.","causal_attribution":"Connects the detections to pay; the amount lost by any individual driver is not reported."},{"id":"c5","status":"reported","evidence":[{"locator":"'but they don’t respond. There’s no room for discussion around the possibility that maybe the camera’s data isn’t clean.'; 'made the appeal process futile.'","relation":"supports","source_id":"s1"}],"assertion":"The Los Angeles driver said requests through his delivery company to see the photos flagged as events went unanswered, and a delivery-company owner said appeals were futile.","causal_attribution":"The drivers' and owner's account of the appeal route."},{"id":"c6","status":"reported","evidence":[{"locator":"'a team of Amazon employees manually reviews all events that are appealed to ensure that erroneous events do not impact drivers or Amazon delivery companies'; 'This technology provides drivers real-time alerts to help them stay safe when they are on the road.'; 'accidents had decreased by 48 percent'","relation":"supports","source_id":"s1"}],"assertion":"Amazon said the cameras give real-time safety alerts, that accidents had fallen 48 percent since installation, and that its employees manually review all appealed events so that erroneous events do not affect drivers.","causal_attribution":"Amazon's response; its safety figures are its own claims."}],"effects":[{"label":"Drivers say false camera events lowered their weekly scores","claim_id":"c2","direction":"negative"},{"label":"Scores decide bonuses and prizes","claim_id":"c4","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.vice.com/en/article/amazons-ai-cameras-are-punishing-drivers-for-mistakes-they-didnt-make/","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-27 (HTTP 200, full article body).","independence_group":"motherboard-netradyne-2021-09-20"}],"version":1,"ai_roles":["institutional_use"],"contexts":["work","finance"],"unknowns":["How many events recorded against these drivers were wrong, and whether any was later overturned on appeal.","How much bonus or prize value any individual driver lost.","Whether Amazon or Netradyne changed the detection thresholds after these reports."],"geography":{"basis":"The drivers worked from Amazon delivery stations in Los Angeles, Sugar Land (Texas), Mobile (Alabama), Oklahoma and Kentucky, and the vans and routes are in the United States.","court_countries":[],"event_countries":["US"],"affected_person_countries":["US"]},"publication":{"basis":"Published because the drivers describe specific detection errors by the AI cameras (cars cutting in scored as unsafe following, mirror checks scored as distraction, a stop sign flagged where none existed) and say those detections lowered the weekly scores that decide their bonuses and prizes. The accounts are attributed, Amazon's response and appeal process are recorded, and the amount lost is left unknown.","reviewed_on":"2026-10-27"},"ai_involvement":{"basis":"Motherboard describes the Netradyne cameras as AI-powered devices that detect driving events (following distance, stop-sign and signal violations, distracted driving) and upload them to a score. The drivers say the cameras' own detections were wrong: a car cutting in was recorded as unsafe following distance, a glance at a mirror as distraction, and a stop sign was flagged on a road that had none. Those detections entered the weekly scores that, according to the drivers and Amazon's internal scorecard document, decide bonuses and prizes.","status":"reported"},"person_relations":["made_claim_about"]},"name":"Amazon delivery drivers say Netradyne AI cameras in their vans flagged driving 'events' they did not commit, lowering the weekly scores that set bonuses","summary":"In September 2021 Motherboard reported accounts from six Amazon delivery drivers in five US states and the owner of an Amazon delivery company, who said the AI-powered Netradyne cameras Amazon had installed in its delivery vans from early 2021 regularly registered driving 'events' that were beyond their control or were not unsafe driving: being cut off by another car, glancing at a side mirror, stopping past a stop sign to see around a bush, or a stop sign flagged on a dirt road with no sign. Each event fed a weekly safe-driving score that, according to the drivers and an internal Amazon document Motherboard obtained, helped decide drivers' bonuses and prizes and delivery companies' bonuses. A driver in Los Angeles said the camera incorrectly penalised him and that requests to see the photos flagged as events went unanswered; a former driver in Texas said events that did not happen affected his score, which decided prizes from his delivery company. Amazon said the technology gives drivers real-time safety alerts, that accidents and violations had fallen since it was installed, and that its staff manually review appealed events so that erroneous ones do not affect drivers. Netradyne did not respond to Motherboard.","incidentDate":"2021-01-01","incidentKind":"ongoing_experience","incidentDatePrecision":"year","exposurePattern":"repeated_interactions","reportedDate":"2021-09-20","aiSystem":"Netradyne AI-powered four-lens cameras installed in Amazon-branded delivery vans, which detect driving 'events' such as following distance, stop-sign and signal violations and distracted driving, upload the footage and feed Amazon's weekly driver and delivery-company scores","aiProduct":"Netradyne cameras","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["professional_harm","financial_loss"],"harmOutcomeSummary":"Drivers told Motherboard that the AI cameras registered events they did not commit or that were not unsafe, lowering the weekly scores that decided their bonuses and prizes; one said winning prizes 'went downhill' once the cameras came, and another described the penalties as upsetting.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":2,"affectedCountStatus":"partial","affectedCountEvidence":"Two drivers are counted: the Los Angeles driver, who said the camera incorrectly penalised him in the data used for his weekly evaluation and bonus, and the former Sugar Land driver, who said events that did not happen affected his score and his prizes. Motherboard spoke to six drivers and a delivery-company owner who described false or unfair events more generally; the others are not counted because a personal score or pay consequence is not reported for each. The fleet size and driver numbers in the article are not counts of harmed people. The drivers are not counted as direct AI users.","victimAgeRange":"adult","jurisdiction":"US","platformType":"other","outcomeStatus":"unknown","primarySourceUrl":"https://www.vice.com/en/article/amazons-ai-cameras-are-punishing-drivers-for-mistakes-they-didnt-make/","primarySourceLabel":"Motherboard (VICE), 20 September 2021: Amazon's AI cameras are punishing drivers for mistakes they didn't make","firstPublishedAt":"2026-10-27T03:09:21.159219+00:00","updatedAt":"2026-10-27T03:09:21.159219+00:00","scopeVersion":"facts-v3","tags":["work","delivery-drivers","workplace-monitoring","in-vehicle-camera","driver-scores","pay","institutional-use"]},{"id":"2024-brasilia-labour-court-orders-ifood-to-compensate-delivery-worker-banned-after-its-face-check-flagged-his-own-photos","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'O aplicativo do iFood exige que, todos os dias, os entregadores enviem uma foto como reconhecimento facial, antes das entregas ou durante o trabalho. O objetivo é evitar fraudes e invasão à conta.'","relation":"supports","source_id":"s1"}],"assertion":"The iFood courier app requires delivery workers to send a photo for facial recognition every day, before deliveries or during work, to prevent fraud and account takeover.","causal_attribution":"Description of the system; no causal claim."},{"id":"c2","status":"reported","evidence":[{"locator":"'O caso ocorreu em 30 de abril de 2024' ... 'apareceu a mensagem dizendo que eu estava banido da plataforma por descumprir as normas' ... 'o motivo do banimento foi por conta de que outra pessoa tentou acessar inúmeras vezes o aplicativo e isso é proibido' ... 'era eu na foto'","relation":"supports","source_id":"s3"},{"locator":"'Pedi que enviassem a foto de quem estava tentando acessar. E para minha surpresa, a pessoa na foto era eu.'","relation":"supports","source_id":"s1"}],"assertion":"The worker says that on 30 April 2024, after completing the facial check, he received a message that he was banned for breaking the rules; iFood later said another person had repeatedly tried to access the account, and the photo it showed was of him.","causal_attribution":"The worker's account (one witness, interviewed by two outlets); accepted as true by the court by default."},{"id":"c3","status":"reported","evidence":[{"locator":"'me tiraram uma renda bem significativa, contava com esse dinheiro toda semana pra pagar minhas contas'","relation":"supports","source_id":"s1"}],"assertion":"The worker says the ban took away a significant income he relied on every week to pay his bills.","causal_attribution":"The worker's account of the consequence."},{"id":"c4","status":"reported","evidence":[{"locator":"'condenou o iFood a pagar R$ 60 mil como indenização por danos morais' ... 'reconheço como verdadeiros: a suspensão do contrato de trabalho por culpa exclusiva da Reclamada; os prejuízos financeiros e psicológicos sofridos pelo Reclamante em razão da impossibilidade de laborar em favor da Reclamada; e a falha racista na ferramenta utilizada para reconhecimento facial' ... 'o iFood não cumpriu o prazo para apresentar a defesa'","relation":"supports","source_id":"s1"},{"locator":"'O iFood não apresentou defesa, segundo o magistrado.'","relation":"supports","source_id":"s2"}],"assertion":"A Brasília labour judge ordered iFood to pay R$60,000 in moral damages, applying a default because iFood did not file a defence in time, and recognised as true the suspension of the contract through iFood's fault, the worker's financial and psychological losses and a racist failure in the facial recognition tool.","causal_attribution":"First-instance judgment as quoted by two outlets (one underlying document); findings rest on a procedural default and are under appeal."},{"id":"c5","status":"reported","evidence":[{"locator":"'O iFood afirmou não existe análise de características raciais no procedimento de reconhecimento facial.'","relation":"supports","source_id":"s1"},{"locator":"'A empresa já apresentou recurso contra a sentença.'","relation":"supports","source_id":"s3"}],"assertion":"iFood says there is no analysis of racial characteristics in its facial recognition procedure, and it has appealed the judgment.","causal_attribution":"iFood's statement and the appeal; contrary evidence on the racial-bias finding."},{"id":"c6","status":"reported","evidence":[{"locator":"'voltou a trabalhar com entrega pelo app'","relation":"supports","source_id":"s1"}],"assertion":"iFood restored the account during the case, and the worker returned to delivering through the app.","causal_attribution":"Outcome for the worker before judgment."},{"id":"c7","status":"reported","evidence":[{"locator":"'Em ambas as fotos está a mesma pessoa: o reclamante (entregador)' ... 'o entregador trabalhava havia três anos para o iFood quando foi bloqueado na plataforma por suposto acesso de outra pessoa na conta'","relation":"supports","source_id":"s2"}],"assertion":"The worker's lawyer argued that both photos iFood cited show the same person, the worker, and that he had worked for iFood for about three years when he was blocked over the supposed access by another person.","causal_attribution":"The worker's pleading as reported by Estadão."}],"effects":[{"label":"delivery worker banned from the app and lost income after its face check flagged his own photos (worker's account; court ruling by default)","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://g1.globo.com/df/distrito-federal/noticia/2025/06/30/justica-condena-ifood-a-indenizar-entregador-do-df-por-danos-morais-apos-falha-racista-em-reconhecimento-facial-entenda.ghtml","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. G1 report by TV Globo Brasília, read in full.","independence_group":"g1-tvglobo-brasilia"},{"id":"s2","url":"https://www.estadao.com.br/politica/coluna-do-estadao/justica-condena-ifood-a-indenizar-entregador-por-falha-racista-em-reconhecimento-facial/","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Estadão column report, read in full; it gives the court as the 8th Labour Court, where G1 gives the 22nd.","independence_group":"estadao-coluna"},{"id":"s3","url":"https://www.terra.com.br/nos/ifood-e-condenado-a-indenizar-entregador-por-falha-racista-em-reconhecimento-facial-no-app,8954fd59fdcfb407081f434e12593ba4gonqqml2.html","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-27 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Terra report based on its own interview with the worker, read in full.","independence_group":"terra-nos"}],"version":1,"ai_roles":["institutional_use"],"contexts":["work","justice"],"unknowns":["Whether any person reviewed the facial check result before the ban, or whether the ban was fully automated.","How long the account stayed blocked before iFood restored it.","Which company supplies the facial recognition technology.","Which Brasília labour court heard the case (G1 says the 22nd, Estadão the 8th).","How the appeal ends."],"geography":{"basis":"G1 describes the worker as a delivery worker of the Federal District (DF) and the ban happened while he was making deliveries there, so the event and the affected person are in Brazil; the case was heard by a labour court in Brasília.","court_countries":["BR"],"event_countries":["BR"],"affected_person_countries":["BR"]},"publication":{"basis":"Published as a contextual case (made_decision_about): a delivery worker says the iFood app's facial recognition check flagged his own photos as another person and the platform banned him, costing income; a Brasília labour court, ruling by default, recognised a racist failure in the facial recognition tool and awarded R$60,000. iFood denies racial analysis and has appealed. No private name appears in the public record.","reviewed_on":"2026-10-27"},"ai_involvement":{"basis":"The iFood app runs a daily facial recognition check on delivery workers' photos (G1). The worker says that after he completed it the platform banned him, iFood support said another person had tried to access his account, and the photos it showed as that person were his own (Terra, G1); his lawyer's case was that both photos show the same person (Estadão). The labour judge, applying a default because iFood filed no defence, recognised as true a racist failure in the tool used for facial recognition (G1, Estadão). The connection between the system's output and the ban is the worker's account, accepted by default; iFood denies racial analysis but no source reports it disputing the ban sequence. Whether a person reviewed the result before the ban is not stated.","status":"reported"},"person_relations":["made_decision_about"]},"name":"Brasília labour court orders iFood to pay delivery worker who says its face check flagged his own photos as another person and got him banned; iFood appeals","summary":"A delivery worker in Brazil's Federal District who had worked for the iFood app for about three years says he was banned from the platform in April 2024 shortly after completing the app's daily facial recognition check. iFood support told him another person had tried to access his account; when he asked to see that person's photos, they were his own. G1 reports that the only difference from his registration photo was his hair, which he was growing out. In a judgment reported in June 2025, a Brasília labour judge, applying a default because iFood had not filed a defence in time, accepted as true that the contract was suspended through iFood's fault, that he suffered financial and psychological losses, and that the facial recognition tool had a racist failure, and ordered iFood to pay R$60,000 in moral damages. iFood denies that its facial recognition procedure analyses racial characteristics and has appealed. The account was restored during the case. No appeal ruling has been reported.","incidentDate":"2024-04-30","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"single_interaction","reportedDate":"2025-06-11","aiSystem":"The facial recognition check in the iFood courier app, which requires delivery workers to submit a photo every day before or during deliveries to prevent fraud and account takeover; the provider of the underlying technology is not named","aiProduct":"iFood facial recognition check","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss","professional_harm","psychological_distress"],"harmOutcomeSummary":"A delivery worker says he was banned from the iFood app after its facial recognition check flagged his own photos as another person and lost income he relied on; a labour court, ruling by default, recognised his financial and psychological losses and awarded R$60,000. iFood denies racial analysis and has appealed.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"exact","affectedCountEvidence":"One delivery worker, the plaintiff. No other affected person is described.","victimAgeRange":"adult","jurisdiction":"BR","platformType":"other","outcomeType":"lawsuit_ongoing","outcomeStatus":"ongoing","primarySourceUrl":"https://g1.globo.com/df/distrito-federal/noticia/2025/06/30/justica-condena-ifood-a-indenizar-entregador-do-df-por-danos-morais-apos-falha-racista-em-reconhecimento-facial-entenda.ghtml","primarySourceLabel":"G1 / TV Globo Brasília, 30 Jun 2025: court orders iFood to compensate delivery worker over facial recognition failure (Portuguese)","firstPublishedAt":"2026-10-27T03:08:51.893173+00:00","updatedAt":"2026-10-27T03:08:51.893173+00:00","scopeVersion":"facts-v3","tags":["facial-recognition","identity-verification","gig-work","delivery-workers","account-ban","labour-court","brazil","made-decision-about"]},{"id":"2024-gelderland-sub-district-judge-used-chatgpt-to-estimate-solar-panel-damages-in-neighbour-dispute-and-rejected-claim","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'Volgens hem is de verminderde opbrengst van de panelen van 77%'; 'Dat leidt volgens hem tot een totale schade van € 21.385,90'","relation":"supports","source_id":"s1"}],"assertion":"The claimant sued his neighbour for about 21,000 euros, saying the neighbour's roof extension cut the yield of his solar panels by 77%.","causal_attribution":"The claimant's case as recorded in the judgment; documented as the content of his claim, not as a fact about the yield."},{"id":"c2","status":"documented","evidence":[{"locator":"'De kantonrechter schat, mede met behulp van ChatGPT, de gemiddelde levensduur van zonnepanelen uit 2009 op 25 à 30 jaar'; 'Uitgaande, wederom mede aan de hand van ChatGPT, van een huidige gemiddelde kWh-prijs van € 0,30'; 'bedraagt het rendementsverlies, met een min of meer gelijkblijvende kWh-prijs, daarmee ongeveer € 2.250,--'","relation":"supports","source_id":"s1"},{"locator":"'Bij de schadeberekening maakt de rechtbank ook gebruik van ChatGPT'","relation":"supports","source_id":"s2"}],"assertion":"The judge estimated the panels' remaining lifespan and the average electricity price partly with the help of ChatGPT and put the yield loss at about 2,250 euros.","causal_attribution":"Stated by the court in its own judgment; the summary relays it. One origin (the judgment)."},{"id":"c3","status":"documented","evidence":[{"locator":"'het maximaal te begroten bedrag aan schade zou hiermee (€ 2.250,-- + € 1.073,40 =) € 3.323,40 bedragen'; 'uiteindelijk aangeboden bedrag van'; 'Er is dus in deze procedure geen grond om'; 'heeft ook geen recht op € 3.300,-, nu hij het aanbod tot het betalen van dit bedrag heeft verworpen'","relation":"supports","source_id":"s1"},{"locator":"'Rechtbank Gelderland oordeelt dat X geen recht heeft op een schadevergoeding in verband met de door Y geplaatste dakopbouw'","relation":"supports","source_id":"s2"}],"assertion":"Because half of the maximum damage of 3,323.40 euros was below the neighbour's 3,300-euro offer, which the claimant had refused, the claim was rejected and he received nothing.","causal_attribution":"The court's reasoning linking the ChatGPT-assisted estimate to the rejection; documented by the judgment."},{"id":"c4","status":"documented","evidence":[{"locator":"'Redelijkerwijs – en ook weer mede aan de hand van ChatGPT – valt aan te nemen dat dit materiaal daardoor niet meer bruikbaar was'; 'wegens schadevergoeding te betalen'; '€ 2.619,02, vermeerderd met de wettelijke rente'","relation":"supports","source_id":"s1"}],"assertion":"The judge also assumed, partly with ChatGPT, that insulation the claimant threw into water was unusable, and on the counterclaim ordered him to pay damages of 2,619.02 euros in total and the costs of the main claim.","causal_attribution":"Stated in the judgment; the ChatGPT-assisted assumption supports only the 204.66-euro insulation item within the total."}],"effects":[{"label":"damages claim rejected after a ChatGPT-assisted damage estimate","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://data.rechtspraak.nl/uitspraken/content?id=ECLI:NL:RBGEL:2024:3636","kind":"court_record","access":"read","language":"nl","translation_note":"Read in Dutch by the research agent (an AI) on 2026-10-24 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Full text of the judgment of 7 June 2024 from the Rechtspraak open-data service (published 26 July 2024); HTTP 200, read in full.","independence_group":"rechtbank-gelderland-judgment-2024-3636"},{"id":"s2","url":"https://www.taxlive.nl/nl/documenten/vn-vandaag/rechter-gebruikt-chatgpt-bij-schadeberekening/","kind":"news_report","access":"read","language":"nl","translation_note":"Read in Dutch by the research agent (an AI) on 2026-10-24 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Taxlive V-N Vandaag summary of the same judgment, edition 12 August 2024; HTTP 200, read in full.","independence_group":"rechtbank-gelderland-judgment-2024-3636"}],"version":1,"ai_roles":["institutional_use"],"contexts":["justice","everyday_life"],"unknowns":["What the judge asked ChatGPT and what it answered.","Whether the claimant appealed and with what result.","How the outcome would have differed without the ChatGPT-assisted estimates; the judgment also rejected the investment claim on other grounds."],"geography":{"basis":"The dispute concerns two attached houses whose owners also live in them, decided by the Rechtbank Gelderland sitting in Nijmegen; the street and town are anonymised in the published judgment, which applies Dutch law and Dutch subsidies.","court_countries":["NL"],"event_countries":["NL"],"affected_person_countries":["NL"]},"publication":{"basis":"Published as an AI-informed decision: in a judgment of 7 June 2024 a Dutch sub-district judge stated that the estimates of a homeowner's solar-panel lifespan and electricity prices were made partly with ChatGPT, and the resulting damage figure led to the rejection of the homeowner's claim against his neighbour. The facts come from the judgment itself.","reviewed_on":"2026-10-24"},"ai_involvement":{"basis":"The judgment states that the judge estimated the lifespan of the claimant's solar panels and the current electricity price partly with the help of ChatGPT, and those figures set the yield loss at about 2,250 euros and the maximum damage below the level at which the neighbour's offer would have been insufficient, leading to the rejection of the claim; the judge also relied partly on ChatGPT to treat insulation material as unusable in the counterclaim. The judgment does not reproduce the questions or answers.","status":"supported"},"person_relations":["informed_decision_about"]},"name":"Dutch sub-district judge used ChatGPT to estimate solar-panel lifespan and power prices in a neighbour dispute, then rejected the homeowner's damages claim","summary":"A homeowner in the Netherlands sued his neighbour, whose permitted roof extension shaded his solar panels, claiming more than 21,000 euros for lost yield and investment. In a judgment of 7 June 2024 (ECLI:NL:RBGEL:2024:3636), the sub-district judge of the Rechtbank Gelderland, sitting in Nijmegen, held that the neighbour had to bear half of the damage and then calculated the damage, estimating the average lifespan of solar panels from 2009 and the current average electricity price partly with the help of ChatGPT, as the judgment states. Using a lifespan of 27.5 years, about 15 remaining years and a price of 0.30 euros per kWh, the judge put the yield loss at about 2,250 euros and the maximum damage at 3,323.40 euros; half of that was less than the 3,300 euros the neighbour had offered, so the claim was rejected and the homeowner, who had refused the offer, received nothing and was ordered to pay costs. On the neighbour's counterclaim, the judge assumed, again partly with ChatGPT, that insulation material the homeowner had thrown into water was no longer usable, and ordered him to pay damages of 2,619.02 euros in total. The judgment does not say what was asked or answered, and whether it was appealed is not reported.","incidentDate":"2024-06-07","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"single_interaction","reportedDate":"2024-07-26","aiSystem":"ChatGPT, which the judgment says was used, together with other reasoning, to estimate the lifespan of solar panels, the average electricity price and whether water-damaged insulation was still usable","aiProduct":"ChatGPT","severity":"low","verificationStatus":"verified","harmCategories":[],"harmOutcomes":["legal_harm","financial_loss"],"harmOutcomeSummary":"A homeowner's damages claim was rejected after the judge calculated the damage partly with ChatGPT's estimates, and the homeowner was ordered to pay costs and counterclaim damages, according to the judgment.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"exact","affectedCountEvidence":"The claimant homeowner, whose claim was rejected in the judgment. The neighbour won the counterclaim and is not counted.","victimAgeRange":"adult","jurisdiction":"NL","platformType":"chatbot","outcomeType":"lawsuit_dismissed","outcomeStatus":"unknown","primarySourceUrl":"https://data.rechtspraak.nl/uitspraken/content?id=ECLI:NL:RBGEL:2024:3636","primarySourceLabel":"Rechtbank Gelderland, judgment of 7 June 2024, ECLI:NL:RBGEL:2024:3636 (Rechtspraak open data)","firstPublishedAt":"2026-10-24T03:15:10.77655+00:00","updatedAt":"2026-10-24T03:15:10.77655+00:00","scopeVersion":"facts-v3","tags":["court","judge-ai-use","chatgpt","damages","netherlands"]},{"id":"2021-telangana-samagra-vedika-algorithm-wrongly-tagged-poor-families-as-car-owners-food-security-cards-denied","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'started arming the officials with Samagra Vedika, an algorithmic system that consolidates citizens'; 'the officials were mandated to check whether the algorithmic system approved the eligibility of the applicant before making their own decision'; 'They could either go with the algorithm'","relation":"supports","source_id":"s1"}],"assertion":"Since 2016 Telangana has used Samagra Vedika, a system that consolidates residents' data from several government databases, to check welfare eligibility; officials processing food security cards were mandated to check whether the system approved an applicant before deciding, and could depart from it only by giving reasons and evidence.","causal_attribution":"Background on the system and how its output entered officials' decisions."},{"id":"c2","status":"reported","evidence":[{"locator":"'Samagra Vedika uses'; 'to uniquely identify a person'; 'If it finds a match close enough violating any of the criteria, it tags the claimant as ineligible'","relation":"supports","source_id":"s1"}],"assertion":"Al Jazeera reports that Samagra Vedika uses machine learning and entity resolution technology to identify a person across databases, and tags a claimant as ineligible when it finds a close enough match violating a criterion such as owning a four-wheeler.","causal_attribution":"The description of the technology is Al Jazeera's, quoting terms whose origin the article does not state; the government describes it as using big data."},{"id":"c3","status":"reported","evidence":[{"locator":"'mistook her late husband'; 'and the authorities accepted the algorithm'; 'finally rejecting it in 2021 because they claimed she possessed a four-wheeler'; 'The authorities did not trust her'; 'they believed their algorithm'","relation":"supports","source_id":"s1"}],"assertion":"According to Al Jazeera, the system mistook a widow's late husband, a rickshaw puller, for a car owner with a similar name; her food security card renewal application was rejected in 2021 because she was said to possess a four-wheeler, and that the authorities believed their algorithm rather than her.","causal_attribution":"Al Jazeera attributes the rejection to the system's wrong match and officials' acceptance of it."},{"id":"c4","status":"reported","evidence":[{"locator":"'found its real owner'; 'the officials agreed there was an algorithmic error but said that her application could not be considered because her total family income exceeded the eligibility limit even though that was not the case'; 'who has been denied ration for more than seven years, is still waiting'","relation":"supports","source_id":"s1"}],"assertion":"After she traced the vehicle's real owner, officials agreed there had been an algorithmic error but said that her application could not be considered because family income exceeded the limit, which Al Jazeera says was not the case; she had been denied the rations for more than seven years when the article was published.","causal_attribution":"The officials' acknowledgement of an algorithmic error is reported by Al Jazeera; the continued denial rests on a different, disputed ground."},{"id":"c5","status":"reported","evidence":[{"locator":"'an algorithm has tagged the family as owning a car, denying them access to subsidised food grains'; 'The family applied for a food security card in 2018 but was rejected in 2021'; 'Instead of three rupees, we spend 10 for every kilogramme of rice'; 'the petitioner is eligible'","relation":"supports","source_id":"s1"}],"assertion":"A second family, whose husband cannot drive, applied in 2018 and was rejected in 2021 for possessing a four-wheeler they do not own; Al Jazeera's caption says an algorithm tagged the family as owning a car, and the woman says the family pays more for the same rice bought from dealers. In November 2023 the High Court of Telangana said the petitioner is eligible whenever the state issues new cards.","causal_attribution":"The algorithmic tag is stated in an Al Jazeera photo caption credited to The Reporters' Collective; the High Court remark is reported, not read in the order itself."},{"id":"c6","status":"reported","evidence":[{"locator":"'The petition lists around 10 cases of wrongful exclusions'; 'Al Jazeera visited three of those and can confirm they were wrongly tagged by Samagra Vedika'","relation":"supports","source_id":"s1"}],"assertion":"Al Jazeera says it visited three of the roughly ten wrongful-exclusion cases listed in a Supreme Court petition and confirmed they were wrongly tagged by Samagra Vedika.","causal_attribution":"Al Jazeera's own verification; the three cases are not identified."},{"id":"c7","status":"disputed","evidence":[{"locator":"'The government has previously said that the technology is of'; 'It has also said that the error of tagging a vehicle or a house to the wrong person happens in less than five percent of all cases'; 'very high levels of accuracy'","relation":"supports","source_id":"s1"}],"assertion":"The Telangana government has said the technology is of high precision, that wrong tagging of a vehicle or house occurs in less than five per cent of cases and has mostly been corrected, and an official said verifications found very high levels of accuracy.","causal_attribution":"Government counter-position; it does not address the individual cases."}],"effects":[{"label":"food security card applications rejected after the system wrongly tagged families as car owners; years without subsidised rations","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.aljazeera.com/economy/2024/1/24/how-an-algorithm-denied-food-to-thousands-of-poor-in-indias-telangana","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-24 (full article). Reported by Tapasya, Kumar Sambhav and Divij Joshi with The Reporters' Collective and the Pulitzer Center's AI Accountability Network; the photo captions are courtesy of The Reporters' Collective.","independence_group":"al-jazeera-reporters-collective-samagra-vedika-investigation"}],"version":1,"ai_roles":["institutional_use"],"contexts":["public_services","finance"],"unknowns":["Which three petition cases Al Jazeera verified, and how many wrongful exclusions the system itself caused.","Whether the widow's card was later issued, and whether the second family has received a card since November 2023.","How the matching works in detail; the state withheld the source code and data formats.","Why the widow's card was first cancelled in 2016."],"geography":{"basis":"Al Jazeera reports from Hyderabad that both families live there and applied to the Telangana government; the Supreme Court of India and the High Court of Telangana heard related petitions.","court_countries":["IN"],"event_countries":["IN"],"affected_person_countries":["IN"]},"publication":{"basis":"Published as a concrete adverse public decision informed by an automated matching system: Al Jazeera reports that Samagra Vedika wrongly matched a widow's late husband to a car owner and tagged another family as owning a car, and that officials rejected their food security cards on those tags; officials later acknowledged an algorithmic error in the widow's case.","reviewed_on":"2026-10-24"},"ai_involvement":{"basis":"Al Jazeera reports that Samagra Vedika matched a widow's late husband to a car owner with a similar name and tagged her family as owning a four-wheeler, that officials (mandated to check the system's approval before deciding) accepted the tag and rejected her food security card in 2021, and that they later agreed there had been an algorithmic error. It describes the system as using machine learning and entity resolution to match people across databases; the government describes it as using big data and disputes the error rate. The source code and data formats were withheld from the reporters.","status":"supported"},"person_relations":["made_claim_about","informed_decision_about"]},"name":"Al Jazeera investigation says Telangana's Samagra Vedika system wrongly matched poor families to car owners, and officials denied them food subsidy cards","summary":"An Al Jazeera investigation published on 24 January 2024, produced with The Reporters' Collective and the Pulitzer Center's AI Accountability Network, reports that Samagra Vedika, a system the Telangana government uses to consolidate residents' data and check welfare eligibility, wrongly tagged poor families in Hyderabad as owning cars, and that officials relied on those tags to deny them food security cards (subsidised rice). Al Jazeera says the system uses machine learning and entity resolution to match people across databases, and that since 2016 officials have been mandated to check whether the system approved an applicant before deciding. In one case it reports that the system mistook a widow's late husband, a rickshaw puller, for a car owner with a similar name; her renewal application was rejected in 2021 for possessing a four-wheeler. After she traced the vehicle's real owner, officials agreed there had been an algorithmic error but said her family income exceeded the limit, which Al Jazeera says was not the case; she had been without the subsidised rations for more than seven years. A second family, whose husband cannot drive, was rejected in 2021 for possessing a four-wheeler; in November 2023 the High Court of Telangana said the petitioner is eligible whenever new cards are issued. The government says the technology is highly precise and that wrong vehicle or house tags occur in under five per cent of cases and have mostly been corrected.","incidentDate":"2021-01-01","incidentKind":"bounded_series","incidentDatePrecision":"year","exposurePattern":"unknown","reportedDate":"2024-01-24","aiSystem":"Samagra Vedika, the Telangana government's data-integration and profiling system developed by Posidex Technologies, which Al Jazeera says uses machine learning and entity resolution technology to identify people across government databases and tag welfare applicants as ineligible","aiProduct":"Samagra Vedika","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss","other_material_harm"],"harmOutcomeSummary":"According to an Al Jazeera investigation, Telangana's Samagra Vedika system wrongly tagged poor families as car owners and officials denied them food security cards on that basis; a widow had been without subsidised rations for over seven years and another family pays market prices for rice.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":2,"affectedCountStatus":"partial","affectedCountEvidence":"Two applicants described with their rejections: the widow and the woman whose family was rejected in 2021 (members of their households are not counted separately). Al Jazeera says it visited three of about ten petition cases and confirmed they were wrongly tagged, without listing them, and gives state re-verification figures that it does not attribute to the system alone; these are not counted.","victimAgeRange":"adult","jurisdiction":"IN","platformType":"other","outcomeType":"media_coverage","outcomeStatus":"ongoing","primarySourceUrl":"https://www.aljazeera.com/economy/2024/1/24/how-an-algorithm-denied-food-to-thousands-of-poor-in-indias-telangana","primarySourceLabel":"Al Jazeera, How an algorithm denied food to thousands of poor in India's Telangana, 24 January 2024","firstPublishedAt":"2026-10-24T03:08:52.234671+00:00","updatedAt":"2026-10-24T03:08:52.234671+00:00","scopeVersion":"facts-v3","tags":["welfare-eligibility-algorithm","entity-resolution","food-security","wrong-match","public-services"]},{"id":"2026-arizona-appeals-court-strikes-self-represented-appellants-ai-hallucinated-citations-orders-fees-sanction","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"In his opening brief with the Court of Appeals, he made eight citations. The court found that six of them were deficient; two were cases that “simply do not exist.”","relation":"supports","source_id":"s1"}],"assertion":"The self-represented appellant's opening brief made eight citations; the court found six deficient, including two cases that do not exist.","causal_attribution":"The court's findings as reported by the Arizona Mirror."},{"id":"c2","status":"reported","evidence":[{"locator":"readily acknowledged that he relied on generative AI to write his brief; he admitted he did nothing to independently verify the citations","relation":"supports","source_id":"s1"}],"assertion":"In a written explanation and at a hearing on the false citations he acknowledged relying on generative AI to write the brief and admitted doing nothing to verify the citations.","causal_attribution":"The appellant's own acknowledgement, recorded in the opinion and reported by the Arizona Mirror."},{"id":"c3","status":"reported","evidence":[{"locator":"Accordingly, citing a hallucinated case in a legal filing is sanctionable conduct; traceable to dealing with the fake citations; appeal was also rejected because there were no supportive legal authorities for his claims once the inaccurate and false citations were stricken","relation":"supports","source_id":"s1"}],"assertion":"The court held that citing a hallucinated case is sanctionable regardless of intent, ordered him to pay the portion of the other party's attorney's fees traceable to the fake citations, and rejected the appeal because no supportive authority remained once the citations were struck.","causal_attribution":"The court's ruling as reported."},{"id":"c4","status":"reported","evidence":[{"locator":"The three-judge panel noted that court rules allow for more severe sanctions, including contempt charges and monetary fines.; isn’t a trained legal professional, it opted to narrowly sanction him","relation":"supports","source_id":"s1"}],"assertion":"The court chose this narrow sanction over contempt or fines because he said he did not intend to deceive and is not a trained legal professional.","causal_attribution":"The court's reasoning as reported."}],"effects":[{"label":"brief citations struck, appeal rejected and fee sanction ordered","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://kiowacountypress.net/content/arizona-court-appeals-citing-fake-ai-generated-cases-can-get-you-sanctioned-no-matter-your","kind":"news","access":"read","language":"en","translation_note":"Read in full in English by the research agent (an AI) on 2026-10-18 (HTTP 200); no translation was needed. Republication of the Arizona Mirror (States Newsroom) report.","independence_group":"arizona-mirror-report-jul-2026"}],"version":1,"ai_roles":["own_use"],"contexts":["justice"],"unknowns":["Which generative AI tool the appellant used.","The amount of the fee award, which depended on the other party's documentation."],"geography":{"basis":"The brief was filed in, and the sanction imposed by, the Arizona Court of Appeals in a Maricopa County probate appeal. Residence is not used.","court_countries":["US"],"event_countries":["US"],"affected_person_countries":[]},"publication":{"basis":"Published on the Arizona Mirror's report of the Court of Appeals opinion of 15 July 2026: a self-represented appellant acknowledged relying on generative AI to write a brief that cited nonexistent cases, and the court struck the citations, rejected the appeal and ordered him to pay the fees they caused. The parties are private persons and are not named.","reviewed_on":"2026-10-18"},"ai_involvement":{"basis":"The appellant acknowledged in a written explanation and at a hearing on the false citations that he relied on generative AI to write his opening brief, which the court found cited two nonexistent cases and misstated others; the AI-generated text reached him (communicated_with), and the court read the fabricated citations, struck them, rejected the appeal for want of authority and imposed a fee sanction for them (informed_decision_about). The tool is not named.","status":"supported"},"person_relations":["communicated_with","informed_decision_about"]},"name":"Arizona appeals court strikes a self-represented appellant's AI-hallucinated citations, rejects his probate appeal and orders him to pay fees as a sanction","summary":"In a published opinion of July 2026, the Arizona Court of Appeals found that a self-represented appellant's opening brief in a probate dispute contained eight citations, six of them deficient, including two cases that do not exist. In a written explanation and at a hearing on the false citations he acknowledged relying on generative AI to write the brief and said he did nothing to verify the citations. The court struck the inaccurate and false citations, rejected the appeal because no supportive authority remained, held that citing a hallucinated case is sanctionable whether or not it is intended, and ordered him to pay the share of the other party's attorney's fees traceable to the fake citations. Because he said he did not intend to deceive and is not legally trained, the court chose this narrow sanction over contempt or fines. It described the opinion as the first published Arizona ruling on generative AI hallucinations in filings.","incidentDate":"2026-07-15","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"single_interaction","reportedDate":"2026-07-19","aiSystem":"Generative AI, not identified in the report, that the appellant said he relied on to write his opening brief, which cited two nonexistent cases and misstated others","aiProduct":"Unidentified AI drafting tool","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm","financial_loss"],"harmOutcomeSummary":"According to the court, the appellant's AI-drafted brief cited nonexistent cases; the citations were struck, the appeal was rejected and he was ordered to pay the share of the other party's attorney's fees caused by the fake citations.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"One person: the self-represented appellant who used generative AI and was sanctioned.","victimAgeRange":"adult","jurisdiction":"US-AZ","platformType":"other","outcomeType":"regulatory_action","outcomeStatus":"resolved","primarySourceUrl":"https://kiowacountypress.net/content/arizona-court-appeals-citing-fake-ai-generated-cases-can-get-you-sanctioned-no-matter-your","primarySourceLabel":"Arizona Mirror (States Newsroom), republished by Kiowa County Press, July 2026: Arizona Court of Appeals on sanctions for fake AI-generated cases","firstPublishedAt":"2026-10-18T03:14:48.762612+00:00","updatedAt":"2026-10-18T03:14:48.762612+00:00","scopeVersion":"facts-v3","tags":["hallucinated-citations","self-represented-litigant","appeal","sanction","attorney-fees","arizona"]},{"id":"2025-shanghai-yangpu-man-used-ai-faked-food-poisoning-medical-records-to-extort-restaurants-convicted","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'要求商家赔偿2000元医疗费，否则将投诉和曝光'; '最终转账赔付'","relation":"supports","source_id":"s1"},{"locator":"'店家赔了钱'","relation":"supports","source_id":"s3"}],"assertion":"A customer told a Yangpu barbecue restaurant he had fallen ill after eating there, produced a hospital visit record and invoice, demanded 2,000 yuan in medical costs under threat of complaints and publicity, and the restaurant paid.","causal_attribution":"Police account relayed by Shangguan; the Procuratorate Daily feature gives the same episode briefly. Different reporting chains on the same police case; marked reported because the details (amount, threat) come from the police account."},{"id":"c2","status":"reported","evidence":[{"locator":"'材料均系伪造'; '利用AI软件对图片进行篡改，替换姓名、调整日期、伪造金额'","relation":"supports","source_id":"s1"},{"locator":"'从未有过相关就医记录'; '于是用手机AI工具生成了虚假材料'","relation":"supports","source_id":"s2"}],"assertion":"Police found the man had no hospital record and the materials were forged; police say he confessed to altering images of other people's medical records with AI software, replacing the name, date and amount.","causal_attribution":"The AI use rests on the man's admission as reported by police (Shangguan) and in China Voice's report of the investigation (Xinhua). The tool is not named, and the accounts differ on whether existing record images were altered or materials generated."},{"id":"c3","status":"reported","evidence":[{"locator":"'成功敲诈勒索两起，未遂两起，非法获利2500元'","relation":"supports","source_id":"s1"},{"locator":"'去年11月第一次索赔成功后，一个月内他又以同样方式对四家餐厅提出索'","relation":"supports","source_id":"s2"}],"assertion":"Police say he succeeded twice and failed twice, obtaining 2,500 yuan; Xinhua reports that after a first successful claim in November of the previous year he made similar claims against four more restaurants within a month.","causal_attribution":"Counts come from the police account (Shangguan) and the China Voice report; they are not reconciled (two plus two versus one plus four), so the record keeps both."},{"id":"c4","status":"reported","evidence":[{"locator":"'上海市杨浦区人民法院审理认定'; '判处拘役四个月、缓刑四个月，并处罚金2000元'","relation":"supports","source_id":"s2"},{"locator":"'利用AI技术伪造就医材料、恶意投诉、敲诈勒索商家的案件依法提起公诉'; '法院审理后以敲诈勒索罪判处'; '利用AI软件生成急性肠胃炎诊断证明和逼真的呕吐物照片'","relation":"supports","source_id":"s3"}],"assertion":"The Yangpu District People's Procuratorate prosecuted him in June 2026, and the Yangpu District People's Court convicted him of extortion and sentenced him to four months' criminal detention, suspended for four months, and a 2,000 yuan fine; the Procuratorate Daily says he used AI software to generate gastroenteritis diagnosis certificates and vomit photos.","causal_attribution":"Two reporting chains give the same sentence; both relay the court outcome and the judgment itself was not read, so the claim is reported."}],"effects":[{"label":"restaurants paid compensation demanded with AI-forged medical records (police: 2,500 yuan from two successful claims)","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.shobserver.cn/sgh/detail?id=1713109","kind":"news_report","access":"read","language":"zh","translation_note":"Read in Chinese and translated by the research agent (an AI) on 2026-10-17; no human reviewer inspected the text. Shangguan News report of the Yangpu public security sub-bureau's account, 26 February 2026.","independence_group":"yangpu-police-account-2026-02"},{"id":"s2","url":"https://app.xinhuanet.com/news/article.html?articleId=20260616b5cfed558eda4c739780064335734766","kind":"news_report","access":"read","language":"zh","translation_note":"Read in Chinese and translated by the research agent (an AI) on 2026-10-17; no human reviewer inspected the text. Xinhua client carry of a China Media Group China Voice report (reporters named in the text), 16 June 2026, drawing on the police investigation and the court judgment.","independence_group":"cmg-china-voice-2026-06"},{"id":"s3","url":"https://www.chinanews.com.cn/fz/2026/10-14/10714250.shtml","kind":"news_report","access":"read","language":"zh","translation_note":"Read in Chinese and translated by the research agent (an AI) on 2026-10-17; no human reviewer inspected the text. China News Service carry of a Procuratorate Daily feature ('来源：检察日报'), 14 October 2026.","independence_group":"procuratorate-daily-2026-10"}],"version":1,"ai_roles":["others_use"],"contexts":["work","justice"],"unknowns":["The AI tool used, and whether it altered existing record images or generated documents and photos from text.","How many restaurants were targeted and paid: the police account gives two successful and two failed claims, Xinhua a first claim plus four more within a month.","The date of the judgment; the text of the judgment was not read."],"geography":{"basis":"The restaurants are in Yangpu district, Shanghai, and the Yangpu police, procuratorate and court handled the case (Shangguan, Xinhua, Procuratorate Daily).","court_countries":["CN"],"event_countries":["CN"],"affected_person_countries":["CN"]},"publication":{"basis":"Published as a contextual case (made_claim_about): Yangpu police say a man used AI software to forge medical records claiming food poisoning and obtained 2,500 yuan from restaurants, which he admitted; the Yangpu court convicted him of extortion. Three reporting chains (police account, China Voice, Procuratorate Daily).","reviewed_on":"2026-10-17"},"ai_involvement":{"basis":"Yangpu police say the man confessed to altering images of other people's medical records with AI software, changing the name, date and amount and adding a beef food-poisoning note, after the hospital confirmed he had no record (Shangguan). China Voice (Xinhua) says he used an AI tool on his phone to generate the false materials, and the Procuratorate Daily says he used AI software to generate diagnosis certificates and vomit photos and was prosecuted for using AI to forge medical materials. The AI output is the forged documents and photos asserting that the restaurants' food had made him ill (made_claim_about), which the restaurants relied on when they paid. The tool is not named.","status":"supported"},"person_relations":["made_claim_about"]},"name":"Shanghai court convicts a man who, police say, used AI software to fake food-poisoning medical records and extort restaurants into paying compensation","summary":"Police in Yangpu district, Shanghai, say a customer extorted restaurants with medical records he had forged using AI software; Xinhua dates his first successful claim to November 2025. In the police account (Shangguan News, 26 February 2026) he claimed to have fallen ill after eating at a barbecue restaurant, showed a hospital visit record and medical invoice, and demanded 2,000 yuan in medical costs, threatening complaints and publicity; the restaurant paid. Police checked with other diners and the hospital, found he had no hospital record and the materials were forged, and say he confessed to using AI software to alter images of other people's medical records, replacing the name, date and amount. Police say he succeeded twice and failed twice, obtaining 2,500 yuan, and that he also filed complaints to market regulators with the forged material to pressure the restaurants. According to Xinhua's June report and the Procuratorate Daily, the Yangpu District People's Court convicted him of extortion and sentenced him to four months' criminal detention, suspended for four months, and a 2,000 yuan fine.","incidentDate":"2025-11-01","incidentKind":"bounded_series","incidentDatePrecision":"month","exposurePattern":"unknown","reportedDate":"2026-02-26","aiSystem":"AI software on the man's phone that, according to Yangpu police, he used to alter images of other people's medical records (name, date, amount) and, according to later reports, to generate gastroenteritis diagnosis certificates, receipts and vomit photos; the tool is not named","aiProduct":"Unidentified image tool","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss"],"harmOutcomeSummary":"According to Yangpu police, restaurants paid compensation demanded with AI-forged medical records: a barbecue restaurant paid 2,000 yuan, and the man obtained 2,500 yuan in total from two successful claims, with two more attempts failing.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"partial","affectedCountEvidence":"Police say two extortions succeeded and two failed. One payer is described: the barbecue restaurant's person in charge, who paid 2,000 yuan (Shangguan). The second paying restaurant's people are not described, so the count is 1 plus an unquantified remainder.","victimAgeRange":"unknown","jurisdiction":"CN","platformType":"other","outcomeType":"criminal_charges","outcomeStatus":"resolved","primarySourceUrl":"https://www.shobserver.cn/sgh/detail?id=1713109","primarySourceLabel":"Shangguan News, 26 February 2026: Yangpu police on a barbecue restaurant that paid 2,000 yuan over forged medical records","firstPublishedAt":"2026-10-17T03:18:51.274802+00:00","updatedAt":"2026-10-17T03:18:51.274802+00:00","scopeVersion":"facts-v3","tags":["fake-medical-records","food-poisoning-claim","extortion","restaurants","criminal-conviction","shanghai","china","made-claim-about"]},{"id":"2026-new-york-federal-court-defaults-client-and-orders-lawyer-to-pay-fees-over-ai-generated-false-citations","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'motion to dismiss that was peppered with false citations'; 'Top alerted the Court to a series of faulty citations'; 'Defendants Top Experience Company LLC (“Top”) and'; 'cited that did not exist, and eight cases that did exist but did not contain the quotes'; 'Once again, his brief contained false citations'","relation":"supports","source_id":"s2"}],"assertion":"In June 2025 the defendant's lawyer filed a brief supporting its motion to dismiss that, a co-defendant's lawyer showed, cited at least thirteen cases that did not exist and eight real cases that did not contain the quoted words; a later response to the court and a proposed reply brief also contained false citations.","causal_attribution":"The court's findings in its sanctions opinion."},{"id":"c2","status":"documented","evidence":[{"locator":"'unreported cases located via Google Scholar, Google search, or the AI-assisted platform vLex, and then cite-checking those cases by running the brief through a different AI program'; 'introduced citation errors that he mistook as corrections and thus neither reviewed nor fixed'; 'I used generative AI to confirm that the information that I wrote was correctly referencing the information that I had'; 'The tool he claimed to use was NotebookLM'; 'The Court does not find the second explanation credible'; 'Court finds that explanation extremely difficult to believe'","relation":"supports","source_id":"s2"}],"assertion":"The lawyer said he located unreported cases through Google Scholar, Google search and the AI-assisted platform vLex, cite-checked the brief by running it through AI programs (vLex or Paxton AI) that he said introduced the errors, and used NotebookLM to review the content of his response; the court did not find his account of how the errors arose credible and found his NotebookLM account extremely difficult to believe.","causal_attribution":"Documented as the lawyer's statements recorded by the court; his account of how the errors arose is his own and the court rejected it in part."},{"id":"c3","status":"documented","evidence":[{"locator":"'by submitting false citations to the Court produced by generative artificial intelligence (“AI”) and then lying about their genesis'; 'had acted in bad faith'","relation":"supports","source_id":"s1"}],"assertion":"The court later described the sanctioned conduct as submitting false citations produced by generative AI and then lying about their origin, and recalled its finding that the lawyer acted in bad faith.","causal_attribution":"The court's own characterisation in the 9 October 2026 order."},{"id":"c4","status":"documented","evidence":[{"locator":"'by striking Affable’s submissions and entering default judgment against it'; 'nor a client from its attorney’s misdeeds'; 'to enter judgment in favor of Plaintiffs and against Affable Avenue LLC'; 'Dated: February 5, 2026'","relation":"supports","source_id":"s2"},{"locator":"'Court imposed terminal sanctions on Affable'","relation":"supports","source_id":"s1"}],"assertion":"On 5 February 2026 the court sanctioned the lawyer by striking his client's motion papers and entering default judgment against the client, holding that a client is not excused from its lawyer's misdeeds.","causal_attribution":"The court's order; the court ties the sanction to the false citations."},{"id":"c5","status":"documented","evidence":[{"locator":"'then-counsel to Defendant Top Experience Company LLC (“Top”), who had provided substantial assistance to the Court by investigating and reporting'; 'awards his firm $38,877.59'; 'for a total award of $38,877.59'; 'must be made by'; 'within 30 days of this Opinion, unless'; 'Dated: October 9, 2026'","relation":"supports","source_id":"s1"}],"assertion":"On 9 October 2026 the court awarded the firm of a co-defendant's former lawyer, who had investigated and reported the false citations, $38,877.59 in fees and costs, payable by the sanctioned lawyer within 30 days.","causal_attribution":"The court's order."}],"effects":[{"label":"client's motion papers struck and default judgment entered against it as a sanction on its lawyer","claim_id":"c4","direction":"negative"},{"label":"lawyer ordered to pay $38,877.59 in fees and costs","claim_id":"c5","direction":"negative"}],"sources":[{"id":"s1","url":"https://storage.courtlistener.com/recap/gov.uscourts.nysd.633195/gov.uscourts.nysd.633195.364.0.pdf","kind":"court_record","access":"read","language":"en","translation_note":"Read in full in English by the research agent (an AI) on 2026-10-16 from CourtListener RECAP storage (HTTP 200, text extracted with pdftotext); no translation was needed. ECF 364, 18 pp.","independence_group":"sdny-24-civ-9429-court-orders"},{"id":"s2","url":"https://storage.courtlistener.com/recap/gov.uscourts.nysd.633195/gov.uscourts.nysd.633195.227.0.pdf","kind":"court_record","access":"read","language":"en","translation_note":"Read in full in English by the research agent (an AI) on 2026-10-16 from CourtListener RECAP storage (HTTP 200, text extracted with pdftotext); no translation was needed. ECF 227, 33 pp.","independence_group":"sdny-24-civ-9429-court-orders"}],"version":1,"ai_roles":["own_use"],"contexts":["justice","work"],"unknowns":["Which tool generated each false citation; the lawyer's account names several tools and the court rejected parts of it.","Whether the lawyer has paid the award, appealed, or been referred to a disciplinary body; neither order mentions a referral.","The amount of damages to be assessed against the defaulted client."],"geography":{"basis":"The filings were made in, and the sanctions imposed by, the U.S. District Court for the Southern District of New York. Where the lawyer and the client are based is not used.","court_countries":["US"],"event_countries":["US"],"affected_person_countries":[]},"publication":{"basis":"Published because the court's two orders, read in full, record the lawyer's own account of using AI tools to find and check citations, describe the false citations as produced by generative AI, and impose default judgment on his client and a $38,877.59 fee award on him because of them.","reviewed_on":"2026-10-16"},"ai_involvement":{"basis":"The lawyer told the court he found cases through the AI-assisted platform vLex and cite-checked his brief by running it through AI programs (vLex or Paxton AI), which he said introduced the errors, and that he used NotebookLM to review a later submission (communicated_with). The court's 9 October 2026 order describes the sanctioned filings as false citations produced by generative AI. The court read those citations and, because of them, struck the client's papers, entered default judgment against the client and ordered the lawyer to pay fees (informed_decision_about). Which tool generated each false citation is not established.","status":"supported"},"person_relations":["communicated_with","informed_decision_about"]},"name":"New York federal court enters default judgment against a defendant and orders its lawyer to pay $38,877 after false citations it found AI produced","summary":"In a federal trademark suit in the U.S. District Court for the Southern District of New York, the court on 5 February 2026 struck a company defendant's motion papers and entered default judgment against it as a sanction on its lawyer, after his filings repeatedly cited nonexistent cases and misattributed quotations. The lawyer said he had found and cite-checked cases with AI tools (vLex and Paxton AI) and used NotebookLM to review a later submission; the court rejected his explanations and found he acted in bad faith. On 9 October 2026 the court described the citations as produced by generative AI and ordered the lawyer to pay $38,877.59 in fees and costs to the firm of a co-defendant's former lawyer, who had reported the false citations to the court.","incidentDate":"2026-02-05","incidentEndDate":"2026-10-09","incidentKind":"bounded_series","incidentDatePrecision":"day","exposurePattern":"repeated_interactions","reportedDate":"2026-02-05","aiSystem":"AI legal research and cite-checking tools the lawyer said he used (vLex with AI functionality, Paxton AI) and NotebookLM, which he said he used to review his response to the court; the court found the citations were produced by generative AI","aiProduct":"vLex, Paxton AI and NotebookLM (reported)","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm","professional_harm","financial_loss"],"harmOutcomeSummary":"According to the court's orders, the defendant's motion papers were struck and default judgment entered against it as a sanction for its lawyer's filings with false citations, which the court describes as produced by generative AI, and the lawyer was ordered to pay $38,877.59 in fees and costs.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"partial","affectedCountEvidence":"One person counted: the lawyer, who described his own use of AI tools and was ordered to pay $38,877.59. The client against which default judgment was entered is a company; the people behind it are not described and are not counted.","victimAgeRange":"adult","jurisdiction":"US-NY","platformType":"other","outcomeStatus":"ongoing","primarySourceUrl":"https://storage.courtlistener.com/recap/gov.uscourts.nysd.633195/gov.uscourts.nysd.633195.364.0.pdf","primarySourceLabel":"U.S. District Court for the Southern District of New York, Opinion and Order on fees, ECF 364, 24 Civ. 9429 (9 October 2026)","firstPublishedAt":"2026-10-16T03:05:26.378079+00:00","updatedAt":"2026-10-16T03:05:26.378079+00:00","scopeVersion":"facts-v3","tags":["united-states","new-york","legal-profession","sanction","hallucinated-citations","default-judgment","attorney-fees"]},{"id":"2026-israel-magistrates-court-strikes-expert-report-resting-on-ai-hallucinated-case-law-costs","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'התרשלה בתפקידה כנאמנת'; 'ביום 10.3.26הגישה הנתבעת 2חוות דעת'","relation":"supports","source_id":"s1"}],"assertion":"The plaintiff alleges that the defendant, a lawyer acting as trustee, negligently transferred estate money to another person's account; the defendant filed a third-party notice against two banks and, on 10 March 2026, a banking expert's opinion.","causal_attribution":"The decision's own account of the pleadings; the negligence claim is an allegation in a continuing suit."},{"id":"c2","status":"documented","evidence":[{"locator":"'מבוססת על פסיקה ,חקיקה וספרות פיקטיביות שהומצאו ע\"י בינה מלאכותית ,ובכלל זה פסק דין שאינו קיים'; 'כאשר שאלת השימוש בבינה מלאכותית מקומה להתברר בחקירה נגדית'","relation":"supports","source_id":"s1"}],"assertion":"One bank asked the court to strike the opinion as resting on case law, legislation and literature invented by artificial intelligence, including a ruling that does not exist; the defendant replied that the question of AI use should be examined in cross-examination.","causal_attribution":"The parties' submissions as recorded by the court."},{"id":"c3","status":"documented","evidence":[{"locator":"'הליך שלא ניתן לאתרו לפי מספר או שם הצדדים'; 'מבדיקה עצמאית לא נמצא כי קיים ספר ,מאמר או ציטוט כזה'; 'מבלי שהוגשה התייחסותו של המומחה לטענות המופנות כנגד חוות דעתו ,ומחדל זה פועל לחובתה של הנתבעת'","relation":"supports","source_id":"s1"}],"assertion":"Checking the sources itself, the court could not locate the cited Supreme Court ruling by number or party names, found no book, article or quotation matching the cited legal text, and found that the quoted statutory sections say something else; the defendant filed no response from the expert.","causal_attribution":"The court's own findings after its own check of the sources."},{"id":"c4","status":"documented","evidence":[{"locator":"'משנמצא כי לב חוות הדעת והנספחים המצורפים אליה נשענים על \"הזיות בינה מלאכותית\" מובהקות ועל עיוות מוחלט של לשון חוק שירותי תשלום'; 'הגשת חוות דעת המבוססת על הזיות של בינה מלאכותית מהווה פגם מהותי המצדיק את הוצאתה המידית מהתיק עוד לפני שלב ההוכחות'; 'ומורה על מחיקת חוות דעתו'","relation":"supports","source_id":"s1"}],"assertion":"The court held that the heart of the opinion and its annexes rest on clear AI hallucinations and a distortion of the statute, that an opinion resting on AI hallucinations is a substantive defect justifying its removal before the evidence stage, and struck it from the file.","causal_attribution":"The court attributes the defects to AI hallucinations; that attribution rests on the invented content, with no admission or named tool."},{"id":"c5","status":"documented","evidence":[{"locator":"'שעה שהתיקון של ההודעה לצד שלישי מבוסס על חוות דעת המומחה מטעם המודיעה ,חוות דעת שמצאתי אותה פגומה והוריתי על מחיקתה מהתיק'; 'דין הבקשה לתיקון ההודעה לצד שלישי להידחות'; 'תישא בהוצאות הבקשות לטובת הצד דים השלישיים בסך של'; '₪ 2,500לכל אחד מהם בנפרד'; 'ניתנה היום ,כ\"ז תשרי תשפ\"ז 08 ,אוקטובר'","relation":"supports","source_id":"s1"}],"assertion":"The court refused the defendant's request to amend the third-party notice, giving as one reason that the amendment rested on the struck opinion, and ordered her to pay costs of 2,500 shekels to each of the two banks within 30 days regardless of the outcome; the decision is dated 8 October 2026.","causal_attribution":"The court's orders. The refusal also rests on independent procedural grounds (no supporting affidavit, lateness, lack of good faith)."}],"effects":[{"label":"defendant's expert opinion struck from the file as resting on AI-hallucinated authorities","claim_id":"c4","direction":"negative"},{"label":"defendant's amendment refused and costs of 2,500 shekels ordered to each of two banks","claim_id":"c5","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.damiencharlotin.com/documents/3278/Avaragl_v._Yaakov_Israel.pdf","kind":"court_record","access":"read","language":"he","translation_note":"Read in full in Hebrew by the research agent (an AI) on 2026-10-16 from the copy of the decision hosted on the AI Hallucination Cases database (HTTP 200, 10 pages, text extracted with pdftotext and bidirectional-control characters removed); the English renderings in this record are the research agent's own translation, with no human translator involved.","independence_group":"nof-hagalil-nazareth-magistrates-court-civil-file-38820-04-24-decision"}],"version":1,"ai_roles":["others_use"],"contexts":["justice","finance"],"unknowns":["Which AI tool, if any, the expert used; the decision names none and records no account from the expert.","Whether the defendant appeals the decision or pays the costs, and how the suit and the third-party notice are decided.","Whether the expert faces any professional consequence."],"geography":{"basis":"The opinion was filed in, and struck by, the Magistrates' Court in Nof HaGalil-Nazareth, Israel. Where the parties live is not stated and is not used.","court_countries":["IL"],"event_countries":["IL"],"affected_person_countries":[]},"publication":{"basis":"Published because the court's decision, read in full, finds that a filed expert opinion rests on AI-hallucinated authorities and distorted statute text, and on that basis strikes it, refuses the amendment built on it and orders the filing defendant to pay costs. The case is an expert witness's report excluded from evidence before trial.","reviewed_on":"2026-10-16"},"ai_involvement":{"basis":"The court's decision finds that the heart of the expert opinion and its annexes rest on clear 'AI hallucinations': a Supreme Court ruling the court could not locate, a cited book it could not find and statutory sections quoted with text the statute does not contain. The court read that output in the filed opinion and, because of it, struck the opinion, refused the amendment that rested on it and ordered the defendant to pay costs (informed_decision_about). The attribution to AI is the court's own finding after checking the sources; no admission, tool or account by the expert is recorded.","status":"supported"},"person_relations":["informed_decision_about"]},"name":"Israeli court strikes a defendant's expert report it found rested on AI-hallucinated case law and orders her to pay the banks' costs","summary":"On 8 October 2026 the Magistrates' Court in Nof HaGalil-Nazareth, Israel, struck from the file a banking expert's opinion that a defendant in a civil negligence suit had filed, finding that the heart of the opinion and its annexes rested on clear 'AI hallucinations': a Supreme Court ruling the court could not locate, a book it could not find and statutory sections quoted with text the statute does not contain. The court also refused the defendant's request to amend her third-party notice against two banks, partly because the amendment relied on the struck opinion, and ordered her to pay costs of 2,500 shekels to each bank. No AI tool is named, and the decision records no response from the expert.","incidentDate":"2026-10-08","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2026-10-08","aiSystem":"Unidentified generative AI tool; the court found that the expert opinion rests on AI hallucinations but names no tool, and the expert did not respond","aiProduct":"Unidentified AI tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm","financial_loss"],"harmOutcomeSummary":"According to the court's decision, the defendant's expert evidence was struck and her amendment refused because the opinion rested on what the court found were AI-hallucinated authorities and distorted statute text, and she was ordered to pay 2,500 shekels in costs to each of two banks.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"exact","affectedCountEvidence":"One person counted: the defendant who filed the opinion, whose evidence was struck, whose amendment was refused and who was ordered to pay costs. The expert is not counted because the decision orders nothing against him; the plaintiff and the banks are not harmed by the ruling.","victimAgeRange":"adult","jurisdiction":"IL","platformType":"other","outcomeStatus":"ongoing","primarySourceUrl":"https://www.damiencharlotin.com/documents/3278/Avaragl_v._Yaakov_Israel.pdf","primarySourceLabel":"Magistrates' Court, Nof HaGalil-Nazareth, civil file 38820-04-24, decision (8 October 2026)","firstPublishedAt":"2026-10-16T03:01:26.843634+00:00","updatedAt":"2026-10-16T03:01:26.843634+00:00","scopeVersion":"facts-v3","tags":["israel","expert-evidence","hallucinated-citations","court-decision","costs","banking"]},{"id":"2025-oregon-federal-court-dismisses-family-plaintiffs-claims-after-briefs-with-ai-fabricated-citations","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'The issue before the Court is the appropriate sanctions for Plaintiffs' misuse of generative artificial intelligence'; 'Plaintiffs filed three briefs containing citations to fifteen non-existent cases and fabricated quotations falsely attributed to eight legitimate authorities'","relation":"supports","source_id":"s1"}],"assertion":"The plaintiffs filed three summary-judgment briefs citing fifteen nonexistent cases and fabricated quotations attributed to eight real authorities, and the court framed its sanctions as being for the plaintiffs' misuse of generative artificial intelligence.","causal_attribution":"The court's own findings."},{"id":"c2","status":"documented","evidence":[{"locator":"'[t]he inclusion of inaccurate citations was inadvertent and the result of reliance on an automated legal citation tool'","relation":"supports","source_id":"s1"}],"assertion":"The plaintiffs told the court the inaccurate citations were inadvertent and resulted from reliance on an automated legal citation tool.","causal_attribution":"The plaintiffs' filing as quoted by the court."},{"id":"c3","status":"documented","evidence":[{"locator":"'The Court finds that Plaintiffs share responsibility with'; 'for the submission of fifteen non-existent cases and eight fabricated quotations'; 'is a serial self-represented litigator who was likely the person responsible for using artificial intelligence to draft the summary judgment briefs in this case'","relation":"supports","source_id":"s1"}],"assertion":"The court found that the plaintiffs shared responsibility with their lead lawyer and that one plaintiff, a serial self-represented litigator, was likely the person responsible for using AI to draft the briefs.","causal_attribution":"A court finding of likelihood on unrebutted circumstantial evidence; the drafter's AI use is not otherwise established."},{"id":"c4","status":"documented","evidence":[{"locator":"'STRICKEN without leave to refile'; 'to pay the Clerk of the Court $15,500 within 30'; 'Defendants are entitled to an award of reasonable attorney' ; 'Plaintiffs' claims are DISMISSED with'; 'Local Rules provides an additional basis for terminating sanctions against Plaintiffs' claims.'","relation":"supports","source_id":"s1"}],"assertion":"The court struck the briefs without leave to refile, ordered the lead lawyer to pay $15,500 to the Clerk, awarded the defendants reasonable fees and dismissed the plaintiffs' claims with prejudice; it found the plaintiffs' continual violations of the Local Rules an additional basis for that dismissal.","causal_attribution":"Established by the order."},{"id":"c5","status":"documented","evidence":[{"locator":"'finds the requested amount of $94,704.38 reasonable'; 'is ORDERED to pay Defendants 15% of that award, $14,205.66'; 'pay Defendants the remaining 85%'; '$80,498.72 within thirty (30) days'","relation":"supports","source_id":"s2"}],"assertion":"In March 2026 the court found fees of $94,704.38 reasonable and ordered local counsel to pay 15% ($14,205.66) and the lead lawyer 85% ($80,498.72).","causal_attribution":"Established by the fees order."},{"id":"c6","status":"documented","evidence":[{"locator":"'Defendants' unopposed Motion for Partial Summary'; 'summary judgment on their eighth counterclaim for breach of contract and ordered specific performance of the inheritance contract'; 'Plaintiffs' Motion for Reconsideration (ECF No. 228) is DENIED.'","relation":"supports","source_id":"s3"}],"assertion":"In June 2026 the court denied the plaintiffs' motion to reconsider its grant of the defendants' unopposed motion for partial summary judgment, which had awarded specific performance of an inheritance contract on a breach-of-contract counterclaim.","causal_attribution":"Established by the reconsideration order; the orders do not say the counterclaim outcome resulted from the AI sanction."}],"effects":[{"label":"family plaintiffs' claims dismissed with prejudice as a sanction for briefs with AI-fabricated citations","claim_id":"c4","direction":"negative"},{"label":"lead lawyer fined $15,500 and two lawyers ordered to pay $94,704.38 in fees","claim_id":"c5","direction":"negative"}],"sources":[{"id":"s1","url":"https://storage.courtlistener.com/recap/gov.uscourts.ord.158388/gov.uscourts.ord.158388.215.0.pdf","kind":"court_record","access":"read","language":"en","translation_note":"Read in full in English by the research agent (an AI) on 2026-10-15 from CourtListener RECAP storage (HTTP 200, text extracted with pdftotext); no translation was needed.","independence_group":"d-oregon-case-1-21-cv-00157-orders"},{"id":"s2","url":"https://storage.courtlistener.com/recap/gov.uscourts.ord.158388/gov.uscourts.ord.158388.225.0.pdf","kind":"court_record","access":"read","language":"en","translation_note":"Read in English by the research agent (an AI) on 2026-10-15 from CourtListener RECAP storage (HTTP 200, pdftotext); the background and fee-allocation sections were inspected.","independence_group":"d-oregon-case-1-21-cv-00157-orders"},{"id":"s3","url":"https://storage.courtlistener.com/recap/gov.uscourts.ord.158388/gov.uscourts.ord.158388.233.0.pdf","kind":"court_record","access":"read","language":"en","translation_note":"Read in English by the research agent (an AI) on 2026-10-15 from CourtListener RECAP storage (HTTP 200, pdftotext); five-page order read in full.","independence_group":"d-oregon-case-1-21-cv-00157-orders"}],"version":1,"ai_roles":["own_use","others_use"],"contexts":["justice","relationships"],"unknowns":["Which AI tool was used; the orders name none.","Who drafted the briefs: the court's finding is one of likelihood, and the plaintiffs gave no explanation.","Whether the fine and fees were paid, and the outcome of the order sent to the Oregon State Bar and of a later show-cause order to the lead lawyer.","Whether the counterclaim ruling was affected by the striking of the plaintiffs' briefs."],"geography":{"basis":"The briefs were filed in and the sanctions imposed by the U.S. District Court for the District of Oregon. Where the briefs were drafted and where the parties live are not used.","court_countries":["US"],"event_countries":["US"],"affected_person_countries":[]},"publication":{"basis":"Published because the court's own orders establish that it dismissed the family plaintiffs' claims with prejudice and sanctioned their lawyers for briefs citing nonexistent cases and fabricated quotations, which the court attributes to the plaintiffs' misuse of generative AI and finds one plaintiff likely drafted with AI. The three orders were read.","reviewed_on":"2026-10-15"},"ai_involvement":{"basis":"The court's 12 December 2025 order describes the sanctions as being for the plaintiffs' misuse of generative artificial intelligence, finds that three briefs cited fifteen nonexistent cases and fabricated quotations, records the plaintiffs' statement that the citations resulted from reliance on an automated legal citation tool, and finds that one plaintiff was likely the person who used AI to draft the briefs (communicated_with). The AI output reached the court in those briefs, and the court dismissed the plaintiffs' claims and sanctioned their lawyers because of it (informed_decision_about). No product is named.","status":"supported"},"person_relations":["communicated_with","informed_decision_about"]},"name":"Oregon federal court dismisses family plaintiffs' claims and fines their lawyer after briefs cited 15 nonexistent cases it attributes to generative AI misuse","summary":"On 12 December 2025 the U.S. District Court for the District of Oregon dismissed with prejudice the claims of plaintiffs in a family dispute over their late mother's estate and family property, as a sanction for what the court called their misuse of generative artificial intelligence, and found their continual violations of the court's Local Rules an additional basis for the dismissal. Three of the plaintiffs' summary-judgment briefs cited fifteen nonexistent cases and fabricated quotations attributed to eight real authorities. The plaintiffs had said the inaccurate citations came from reliance on an automated legal citation tool. The court found that the plaintiffs shared responsibility with their lead lawyer and that one plaintiff, a frequent self-represented litigant, was likely the person who used AI to draft the briefs. It struck the briefs, ordered the lead lawyer to pay $15,500 to the court and awarded the defendants their fees. In March 2026 it set those fees at $94,704.38, ordering the lead lawyer to pay 85% and local counsel 15%. In June 2026 it denied the plaintiffs' motion to reconsider a later ruling for the defendants on a counterclaim.","incidentDate":"2025-01-31","incidentEndDate":"2025-12-12","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2025-12-12","aiSystem":"Generative AI used to draft summary-judgment briefs, according to the court; the plaintiffs described it as an automated legal citation tool. No product is named in the orders","aiProduct":"Unidentified AI drafting tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm","professional_harm","financial_loss"],"harmOutcomeSummary":"According to the court's orders, the plaintiffs' claims were dismissed with prejudice because their briefs cited nonexistent cases and fabricated quotations that the court attributes to misuse of generative AI, which it found one plaintiff likely used to draft them; the lead lawyer was fined $15,500 and the two lawyers were ordered to pay $94,704.38 in fees.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":0,"otherPeopleHarmedMin":3,"affectedCountStatus":"partial","affectedCountEvidence":"Three people counted from the orders: the individual plaintiff whose claims were dismissed (found likely to have drafted the briefs with AI, so that AI use is a finding of likelihood and she is not counted as an established user), the lead lawyer (fined and ordered to pay 85% of fees) and local counsel (ordered to pay 15% of fees). The estate whose claim was also dismissed has beneficiaries the orders do not count; the LLC plaintiff is not a person.","victimAgeRange":"adult","jurisdiction":"US-OR","platformType":"other","outcomeType":"lawsuit_dismissed","outcomeStatus":"resolved","primarySourceUrl":"https://storage.courtlistener.com/recap/gov.uscourts.ord.158388/gov.uscourts.ord.158388.215.0.pdf","primarySourceLabel":"U.S. District Court for the District of Oregon, Opinion and Order, ECF 215, case 1:21-cv-00157 (12 December 2025)","firstPublishedAt":"2026-10-15T03:05:45.14301+00:00","updatedAt":"2026-10-15T03:05:45.14301+00:00","scopeVersion":"facts-v3","tags":["united-states","oregon","legal-profession","sanction","hallucinated-citations","claims-dismissed","client-responsibility"]},{"id":"2026-alabama-supreme-court-dismisses-family-trust-appeal-over-lawyer-ai-hallucinated-citations","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'filed briefs in this appeal that contain numerous invalid, inaccurate, and/or irrelevant citations to legal authorities'; 'They appear to be artificial-intelligence (\"AI\") \"hallucinations,\" i.e., fake authorities created by an AI system'; 'Most of the legal authorities in the briefs are misquoted or nonexistent'","relation":"supports","source_id":"s1"}],"assertion":"The appellants' lawyer filed briefs whose citations the court found numerous, invalid, inaccurate or irrelevant, most of them misquoted or nonexistent, and which the court said appear to be AI hallucinations.","causal_attribution":"The court's own findings."},{"id":"c2","status":"documented","evidence":[{"locator":"'first use of an AI research tool that summarized commentary not readily available through standard legal databases'","relation":"supports","source_id":"s1"}],"assertion":"In a reply-brief footnote quoted by the court, the lawyer attributed two misquotations to an AI research tool that summarised commentary.","causal_attribution":"The lawyer's statement as quoted in the opinion."},{"id":"c3","status":"documented","evidence":[{"locator":"'Hall admitted that many of the legal authorities in his briefs do not exist and that he improperly used AI to obtain those authorities'; 'Hall also informed the Court that he had self-reported his actions to the Alabama State Bar'","relation":"supports","source_id":"s1"}],"assertion":"At the 4 February 2026 show-cause hearing the lawyer admitted that many authorities in his briefs did not exist and that he had improperly used AI to obtain them, and said he had self-reported to the Alabama State Bar.","causal_attribution":"The court's record of the hearing."},{"id":"c4","status":"documented","evidence":[{"locator":"'Because those briefs are grossly deficient under the Alabama Rules of Appellate Procedure, we dismiss the appeal'; 'the dismissal of this appeal is imposed as a sanction because of the particularly egregious facts before us'","relation":"supports","source_id":"s1"}],"assertion":"The court dismissed the appeal because the briefs were grossly deficient, and said the dismissal was imposed as a sanction because of the particularly egregious facts.","causal_attribution":"Established by the opinion."},{"id":"c5","status":"documented","evidence":[{"locator":"'Hall is hereby ordered to pay attorney fees and costs'; 'in the amount of $17,200'; 'Hall is ordered to pay double costs'; 'We also refer Hall to the Alabama State Bar for potential discipline'","relation":"supports","source_id":"s1"}],"assertion":"The court ordered the lawyer to pay $17,200 in attorney fees and costs and double costs of the appeal, and referred him to the Alabama State Bar for potential discipline.","causal_attribution":"Established by the opinion."},{"id":"c6","status":"documented","evidence":[{"locator":"'SELLERS, Justice (concurring in part and dissenting in part)'; 'I do not agree that dismissal of the appeal is appropriate as a sanction for the conduct of a lawyer'; 'I would affirm the judgment of the Mobile Circuit'; 'MENDHEIM, Justice (concurring in part and dissenting in part)'; 'I concur in the dismissal of this appeal'; 'His representation on appeal was of no benefit to them'","relation":"supports","source_id":"s1"}],"assertion":"Justice Sellers, concurring in part and dissenting in part, disagreed that dismissal was an appropriate sanction for a lawyer's conduct but would have affirmed the trial court's judgment on the merits; Justice Mendheim, who concurred in the dismissal, wrote that the lawyer's representation on appeal was of no benefit to the clients.","causal_attribution":"Two separate opinions in the same record; neither is the court's holding."}],"effects":[{"label":"two appellants' appeal was dismissed as a sanction because their lawyer's briefs relied on AI-obtained authorities that were fake or misquoted","claim_id":"c4","direction":"negative"},{"label":"the lawyer was ordered to pay $17,200 and double costs, barred from filing alone in the court and referred to the state bar","claim_id":"c5","direction":"negative"}],"sources":[{"id":"s1","url":"https://publicportal-api.alappeals.gov/courts/68f021c4-6a44-4735-9a76-5360b2e8af13/cms/case/47683AFC-2F85-469D-8562-9DCA094BB680/docketentrydocuments/D95CDD15-7428-4CC9-8389-3CC27AC88089","kind":"court_record","access":"read","language":"en","translation_note":"Read in full in English by the research agent (an AI) on 2026-10-14 from CourtListener's archived copy of this court PDF (storage.courtlistener.com, HTTP 200, text extracted with pdftotext) because the court portal URL returned HTTP 403 to the agent; no translation was needed.","independence_group":"alabama-supreme-court-sc-2025-0106"}],"version":1,"ai_roles":["own_use","others_use"],"contexts":["justice","work","relationships"],"unknowns":["Which AI tool the lawyer used; the opinion names none.","Whether the dismissal changed the clients' substantive position; the trial court's summary judgment stands, the main opinion does not decide the merits, and one justice would have affirmed the judgment.","The outcome of the Alabama State Bar referral."],"geography":{"basis":"The appeal was decided by the Supreme Court of Alabama on appeal from the Mobile Circuit Court. The appellants' residence is not used.","court_countries":["US"],"event_countries":["US"],"affected_person_countries":[]},"publication":{"basis":"Published because the Supreme Court of Alabama's opinion establishes that it dismissed two appellants' appeal as a sanction after their lawyer filed briefs built on fabricated authorities that he admitted he had obtained by improperly using AI, and that it sanctioned the lawyer and referred him to the state bar. The opinion was read in full.","reviewed_on":"2026-10-14"},"ai_involvement":{"basis":"The court found the briefs' authorities appeared to be AI hallucinations, the lawyer admitted at the show-cause hearing that he had improperly used AI to obtain authorities that do not exist, and his reply-brief footnote quoted by the court says an AI research tool misattributed quotations. The AI output reached the court through the briefs, and the court dismissed the clients' appeal because those briefs rested on fabricated authorities (informed_decision_about for the clients); the lawyer received the fabricated output from the tool (communicated_with) and was sanctioned.","status":"supported"},"person_relations":["communicated_with","informed_decision_about"]},"name":"Alabama Supreme Court dismisses two siblings' family-trust appeal as a sanction after their lawyer admits using AI that produced fake citations","summary":"On 24 April 2026 the Supreme Court of Alabama dismissed an appeal by two siblings in a family dispute over living trusts because their lawyer, W. Perry Hall, had filed briefs in which most of the legal authorities were misquoted or did not exist. The court said they appeared to be AI hallucinations, and at a show-cause hearing in February 2026 Hall admitted that many authorities did not exist and that he had improperly used AI to obtain them. The court ordered Hall to pay the opposing party's fees and costs of $17,200 and double appeal costs, barred him from filing in the court without a co-signing attorney and referred him to the Alabama State Bar. It said the dismissal was imposed as a sanction because of the particularly egregious facts. One justice, concurring in part and dissenting in part, objected to dismissing the appeal as a sanction for the lawyer's conduct but would have affirmed the judgment against the clients on the merits.","incidentDate":"2026-04-24","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2026-04-24","aiSystem":"An AI research tool that, according to the lawyer's reply-brief footnote quoted by the court, summarised legal commentary and misattributed quotations; the lawyer admitted using AI to obtain authorities that do not exist. No product is named in the opinion","aiProduct":"Unidentified AI research tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm","professional_harm","financial_loss"],"harmOutcomeSummary":"According to the Supreme Court of Alabama, the siblings' appeal was dismissed as a sanction because their lawyer's briefs relied on authorities that were misquoted or did not exist, which he admitted he had obtained by improperly using AI; the lawyer was ordered to pay $17,200 in fees and costs plus double appeal costs and was referred to the state bar.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":1,"otherPeopleHarmedMin":2,"affectedCountStatus":"exact","affectedCountEvidence":"One AI user: the appellate lawyer, sanctioned and referred to the bar. Two other people: the two appellants whose appeal was dismissed. The appellee, whose fees were reimbursed, is not counted.","victimAgeRange":"adult","jurisdiction":"US-AL","platformType":"other","outcomeType":"regulatory_action","outcomeStatus":"resolved","primarySourceUrl":"https://publicportal-api.alappeals.gov/courts/68f021c4-6a44-4735-9a76-5360b2e8af13/cms/case/47683AFC-2F85-469D-8562-9DCA094BB680/docketentrydocuments/D95CDD15-7428-4CC9-8389-3CC27AC88089","primarySourceLabel":"Supreme Court of Alabama, opinion in SC-2025-0106 (released 24 April 2026)","firstPublishedAt":"2026-10-14T03:11:52.987847+00:00","updatedAt":"2026-10-14T03:11:52.987847+00:00","scopeVersion":"facts-v3","tags":["united-states","alabama","legal-profession","sanction","hallucinated-citations","appeal-dismissed"]},{"id":"2025-quebec-appeal-court-costs-self-represented-appellant-ai-drafted-filings-nonexistent-judgments","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'Elle réclame aussi 10 000 $ en dommages-intérêts'; 'références juridiques inexistantes générées par l’intelligence artificielle dans ses actes de procédure'","relation":"supports","source_id":"s1"}],"assertion":"The respondent asked the Court of Appeal for $10,000 in damages, alleging that the appellant had included non-existent legal references generated by artificial intelligence in his filings.","causal_attribution":"The decision documents that the respondent made this request; the allegation is hers."},{"id":"c2","status":"documented","evidence":[{"locator":"'parce qu’il a utilisé l’intelligence artificielle pour la rédaction'; 'déclaration d’appel et de son exposé, qui comportent deux références à des arrêts inexistants et une référence inexacte quant à un paragraphe cité'; 'L’appelant a contrevenu à cet Avis de la Juge en chef'; 'une forme de négligence plutôt que de la mauvaise foi'","relation":"supports","source_id":"s1"}],"assertion":"Ruling on the respondent's request, the court described the appellant's notice of appeal and brief as written with artificial intelligence and as containing two references to non-existent judgments and one inaccurate reference to a cited paragraph, and found that he had breached the Chief Justice's notice on AI use through negligence rather than bad faith.","causal_attribution":"Finding stated in the Court of Appeal decision of 1 October 2026; the decision does not name the tool."},{"id":"c3","status":"documented","evidence":[{"locator":"'une telle connaissance de la part de l’appelant n’a pas été démontrée'; 'il y a lieu de le condamner aux frais de justice en vertu de l’article 340 C.p.c.'; 'en matière familiale, de chacune'; 'REJETTE la demande visant à sanctionner un manquement au déroulement'; 'LE TOUT, avec les frais de justice en faveur de l’intimée.'","relation":"supports","source_id":"s1"}],"assertion":"Because knowledge of the invented references was not shown, the court rejected the separate monetary sanction but ordered court costs against the appellant under article 340 C.p.c. to sanction the conduct.","causal_attribution":"Disposition stated in the decision; the costs are the court's response to the AI-drafted filings."},{"id":"c4","status":"documented","evidence":[{"locator":"'ACCUEILLE en partie l’appel'; 'peine de 100 heures de travail d’utilité sociale'; 'une mesure coercitive d’emprisonnement de 30 jours'","relation":"supports","source_id":"s1"}],"assertion":"On the merits the court partly allowed the appeal for reasons unrelated to AI and reduced the contempt penalty to 100 hours of community work and 30 days of coercive imprisonment.","causal_attribution":"Disposition stated in the decision; the contempt arose from non-compliance with family-court orders, not from AI use."}],"effects":[{"label":"a self-represented appellant was ordered to pay court costs because his AI-drafted appeal filings cited non-existent judgments","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.damiencharlotin.com/documents/3266/Droit-de-la-famille-261553-2026-QCCA-1300.pdf","kind":"court_record","access":"read","language":"fr","translation_note":"Read in full in French by the research agent (an AI) on 2026-10-12 from the copy of the decision hosted by the AI Hallucination Cases database (PDF, HTTP 200, text extracted with pdftotext); the English assertions are the agent's translation and no human translator or reviewer was involved. The locators quote the French original.","independence_group":"qcca-2026-qcca-1300-decision"}],"version":1,"ai_roles":["own_use"],"contexts":["justice","relationships"],"unknowns":["Which AI tool the appellant used and what it produced beyond the two non-existent judgments and the inaccurate paragraph reference.","The amount of the court costs ordered against the appellant.","Whether the appellant has complied with the underlying orders or served the coercive imprisonment."],"geography":{"basis":"The appeal was heard and decided by the Quebec Court of Appeal from a Quebec Superior Court judgment (decision caption and paragraph 1), and the filings were made in that court. The decision states no residence for the appellant, so affected_person_countries is left empty.","court_countries":["CA"],"event_countries":["CA"],"affected_person_countries":[]},"publication":{"basis":"Published on the court's own decision: the Quebec Court of Appeal described a self-represented appellant's appeal filings as written with AI and as citing non-existent judgments, held that he had breached its AI notice and ordered costs against him. The facts are documented by the decision; the appellant is a private party in a family matter and is not identified.","reviewed_on":"2026-10-12"},"ai_involvement":{"basis":"Ruling on the respondent's request (paragraph 28), the Quebec Court of Appeal decision describes the appellant's notice of appeal and brief as written with artificial intelligence and as containing two references to non-existent judgments and an inaccurate paragraph reference. The court treated that AI drafting and those references as established: it held that he had breached its AI notice (paragraph 30), contrasted his lack of shown knowledge of the hallucinations with an earlier case (paragraphs 31-32) and ordered costs against him for that reason. The AI output reached him as drafted text he filed (communicated_with) and informed the court's costs decision about him (informed_decision_about). The decision does not identify the tool or quote its output.","status":"supported"},"person_relations":["communicated_with","informed_decision_about"]},"name":"Quebec Court of Appeal orders costs against a self-represented appellant whose AI-drafted filings cited two non-existent judgments (1 Oct 2026)","summary":"On 1 October 2026 the Quebec Court of Appeal ordered court costs against a self-represented appellant in a family-law contempt appeal as a sanction for appeal filings that, in the court's description, he had written with artificial intelligence and that contained two references to non-existent judgments and one inaccurate reference to a cited paragraph. The court held that AI may be used to draft appeal filings only if the litigant verifies the authorities, found that he had breached the Chief Justice's notice on AI use, and described his conduct as negligence rather than bad faith. It rejected the respondent's request for a separate $10,000 sanction and, for reasons unrelated to AI, reduced the contempt penalty imposed at first instance. The decision does not name the AI tool.","incidentDate":"2025-12-15","incidentEndDate":"2026-10-01","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"single_interaction","reportedDate":"2026-10-01","aiSystem":"Artificial intelligence, not identified in the decision, which the self-represented appellant used to draft his notice of appeal and brief","aiProduct":"Unidentified AI drafting tool","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm"],"harmOutcomeSummary":"The Quebec Court of Appeal ordered court costs against the self-represented appellant as a sanction for filings he drafted with AI that cited non-existent judgments, holding that he breached the court's AI notice through negligence (2026 QCCA 1300, 1 October 2026). The amount of the costs is not stated.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"One person: the self-represented appellant who used AI to draft the filings and against whom costs were ordered. The respondent and other family members are not counted as harmed by the AI-related sanction.","victimAgeRange":"adult","jurisdiction":"CA","platformType":"other","outcomeType":"regulatory_action","outcomeStatus":"resolved","primarySourceUrl":"https://www.damiencharlotin.com/documents/3266/Droit-de-la-famille-261553-2026-QCCA-1300.pdf","primarySourceLabel":"Quebec Court of Appeal, Droit de la famille — 261553, 2026 QCCA 1300, 1 October 2026 (copy hosted by the AI Hallucination Cases database)","firstPublishedAt":"2026-10-12T03:17:20.356128+00:00","updatedAt":"2026-10-12T03:17:20.356128+00:00","scopeVersion":"facts-v3","tags":["hallucinated_citations","courts","self_represented_litigant","costs_order","family_law","canada","quebec"]},{"id":"2026-south-africa-sassa-elife-certification-portal-facial-verification-failures-reported-by-pensioners","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"The eLife Certification was implemented on March 30, 2026, as a digital verification tool to curb fraud.","relation":"supports","source_id":"s1"},{"locator":"Beneficiaries who fail to complete the process risk having their grants suspended.","relation":"supports","source_id":"s1"},{"locator":"secure biometric verification through the electronic Know Your Client (eKYC) system","relation":"supports","source_id":"s2"},{"locator":"Beneficiaries who fail to complete life certification as directed may face payment delays or suspension of their grants.","relation":"supports","source_id":"s2"}],"assertion":"SASSA's eLife Certification is a self-service life certification on its online portal that uses biometric verification through its electronic Know Your Client (eKYC) system. IOL reports it was implemented on 30 March 2026. SASSA said beneficiaries who fail to complete life certification as directed may face payment delays or suspension of their grants.","causal_attribution":"The implementation date comes from IOL alone. SASSA statements read do not give a start date. The suspension consequence is SASSA's own stated rule and no suspension of either pensioner's grant is reported."},{"id":"c2","status":"reported","evidence":[{"locator":"SASSA would like to apologise to all our beneficiaries who could not access our self-service portal after they were notified to undertake the eLife Certification verification process by the Agency.","relation":"supports","source_id":"s2"},{"locator":"there were system glitches which led to delay and disruptions in completing eLife Certification, leading to long queues at SASSA offices.","relation":"supports","source_id":"s2"},{"locator":"the Agency can report that the challenge has been resolved.","relation":"supports","source_id":"s2"}],"assertion":"On 10 April 2026 SASSA apologised to beneficiaries who could not access its self-service portal after being notified to complete eLife Certification. It said system glitches linked to interfaces with other departments had caused delays and disruptions and long queues at its offices, and said the challenge had been resolved.","causal_attribution":"SASSA's own account of its portal. The statement does not mention facial recognition and attributes the glitches to interfaces with other departments."},{"id":"c3","status":"reported","evidence":[{"locator":"both pensioners, told IOL they had tried to complete the facial recognition option on the portal but to no avail.","relation":"supports","source_id":"s1"},{"locator":"We have been trying since Thursday, April 2, to do the selfie bit but it does not work after trying for 22 times","relation":"supports","source_id":"s1"}],"assertion":"A pensioner couple told IOL they had tried the facial recognition option on the eLife portal 22 times since 2 April 2026 without success. IOL published the account on 23 April 2026.","causal_attribution":"Single first-person account relayed by one outlet. The cause of the failures is not established. The department attributes facial verification problems in general to poor lighting, unstable connectivity or missing Home Affairs biometric records (see c6)."},{"id":"c4","status":"reported","evidence":[{"locator":"IOL has been inundated with emails and calls from beneficiaries nationwide claiming they were unable to complete the mandatory certification process.","relation":"supports","source_id":"s1"},{"locator":"beneficiaries cite consistent failures with facial recognition and one-time pins (OTPs).","relation":"supports","source_id":"s1"},{"locator":"The system just kept on loading and never went through to my profile. I have been struggling for three weeks","relation":"supports","source_id":"s1"},{"locator":"Then also indicated that the Department of Home Affairs is not available to verify my particulars","relation":"supports","source_id":"s1"}],"assertion":"IOL reported that beneficiaries nationwide contacted it to say they could not complete the certification and cited failures with facial recognition and one-time PINs. One pensioner said the site kept loading without reaching their profile for three weeks, and reported a one-time PIN rejection and a message that Home Affairs was not available to verify their particulars.","causal_attribution":"Unquantified accounts relayed by IOL. The pensioner who described the loading failure and the one-time PIN rejection did not describe a facial recognition failure."},{"id":"c5","status":"reported","evidence":[{"locator":"The Sassa portals are working as evidenced by the number of clients who have accessed the online services.","relation":"supports","source_id":"s1"},{"locator":"As of April 16, 2026, 13,644 (88%) of the 15,499 unique clients who accessed the online verification services via the client portal were successfully verified","relation":"supports","source_id":"s1"},{"locator":"However, Sassa admitted the system has been working intermittently.","relation":"supports","source_id":"s1"}],"assertion":"A SASSA spokesperson told IOL the portals are working and that, as of 16 April 2026, 13,644 (88%) of 15,499 unique clients who accessed the online verification services via the client portal were successfully verified. IOL reports SASSA admitted the system had been working intermittently.","causal_attribution":"SASSA's own figures and account, which were not independently verified. The figures count clients who accessed online verification and do not report how many failed the facial step."},{"id":"c6","status":"reported","evidence":[{"locator":"SASSA said most non-verification cases were driven by beneficiaries failing to respond to notifications, not completing life certification, or unsuccessful facial recognition attempts on online platforms.","relation":"supports","source_id":"s3"},{"locator":"In such cases, beneficiaries are redirected to fingerprint biometric verification at local offices.","relation":"supports","source_id":"s3"},{"locator":"The agency said it had recorded 7,779 complaints linked to its electronic facial biometric system","relation":"supports","source_id":"s3"},{"locator":"The department attributed facial verification issues to poor lighting, unstable connectivity, or missing biometric records at the Department of Home Affairs.","relation":"supports","source_id":"s3"},{"locator":"The agency said it had recorded 7 779 complaints linked to its electronic facial biometric system.","relation":"supports","source_id":"s4"}],"assertion":"IOL's 24 May 2026 report on a parliamentary reply says SASSA said most non-verification cases were driven by beneficiaries who did not respond to notifications, did not complete life certification, or had unsuccessful facial recognition attempts on online platforms, and that such beneficiaries are redirected to fingerprint verification at local offices. IOL says SASSA had recorded 7,779 complaints linked to its electronic facial biometric system and that the department attributed facial verification issues to poor lighting, unstable connectivity or missing biometric records at the Department of Home Affairs.","causal_attribution":"SASSA's account relayed by two outlets that share one reporter and one reply, so they are one chain. The reports do not date the complaints or say whether they concern eLife Certification or earlier online facial verification, and do not say how many were repeat complaints."},{"id":"c7","status":"reported","evidence":[{"locator":"with 67,868 grants suspended in the third quarter alone","relation":"supports","source_id":"s3"},{"locator":"a grant is suspended 2 months after a beneficiary was notified to conduct a review and has not yet done so","relation":"supports","source_id":"s3"},{"locator":"around 70 000 grants have been suspended due to beneficiaries failing to come forward for review","relation":"supports","source_id":"s5"}],"assertion":"The parliamentary reply reported 67,868 grants suspended in the third quarter (the reports do not identify the period) under a rule that a grant is suspended two months after a beneficiary was notified to conduct a review and has not yet done so. A Parliament committee statement of 5 February 2026 reported around 70,000 grants suspended for beneficiaries failing to come forward for review.","causal_attribution":"The suspension figures are context. No inspected report says how many suspensions followed a facial verification failure, and the committee statement predates the eLife implementation date reported by IOL."}],"effects":[{"label":"Pensioners reported repeated failures of the facial recognition step in a mandatory online life certification, with a stated risk of payment delay or grant suspension if it is not completed","claim_id":"c3","direction":"negative"},{"label":"SASSA reported delays, disruptions and long queues at its offices after beneficiaries could not use the self-service certification portal, which SASSA attributed to interfaces with other departments (its statement does not mention facial recognition)","claim_id":"c2","direction":"negative"}],"sources":[{"id":"s1","url":"https://iol.co.za/news/south-africa/2026-04-23-sassa-elife-certification-portal-beneficiaries-report-ongoing-glitches-despite-agency-denials/","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"iol-elife-glitches-report"},{"id":"s2","url":"https://www.gov.za/news/media-statements/sassa-self-service-system-restored-following-earlier-challenges-10-apr-2026","kind":"official_statement","access":"read","language":"en","translation_note":"","independence_group":"sassa-10-apr-2026-statement"},{"id":"s3","url":"https://iol.co.za/news/south-africa/2026-05-24-sassas-biometric-rollout-leaves-thousands-without-grants-amid-fraud-crackdown/","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"parliamentary-reply-iol-report"},{"id":"s4","url":"https://dailyvoice.co.za/news/2026-05-25-sassas-face-palm-facial-recognition-tech-linked-to-the-suspension-of-68-000-grants/","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"parliamentary-reply-iol-report"},{"id":"s5","url":"https://www.parliament.gov.za/index.php/press-releases/media-statement-committee-notes-grant-reviews-are-necessary-protect-poor-and-public-funds","kind":"official_statement","access":"read","language":"en","translation_note":"","independence_group":"parliament-committee-5-feb-2026-statement"}],"version":1,"ai_roles":["institutional_use"],"contexts":["public_services","accessibility"],"unknowns":["The reports do not say how many of the 67,868 grant suspensions (third quarter, period not identified) followed unsuccessful facial recognition. The suspensions are governed by a review non-response rule, and a committee statement dated 5 February 2026 already reported around 70,000 suspensions before the eLife implementation date reported by IOL.","The 30 March 2026 implementation date comes from IOL only. The SASSA statements read do not give a start date.","SASSA's statement refers to facial recognition on online platforms in general. Neither IOL 24 May 2026 nor Daily Voice 25 May 2026 mentions eLife Certification, and IOL places the reply within a biometric verification rollout dated from September 2025 (compulsory biometric enrolment for new applicants, verified at local offices). Whether the 7,779 complaints and the redirections to fingerprint checks concern eLife Certification at all is not stated. Facial photo verification through the same eKYC system for Social Relief of Distress grant applicants was reported earlier (Biometric Update, August 2024, and Corruption Watch, January 2025) and is a separate earlier deployment that this case does not cover.","Whether the pensioner couple later completed certification, or whether any grant was delayed or suspended, is not reported.","The cause of the couple's failed attempts is not established (lighting, connectivity, missing Home Affairs records, or the facial matching itself).","SASSA said on 10 April 2026 that the problem was resolved, and IOL reported failures continuing on 23 April 2026. The 88% success figure is SASSA's and was not checked.","The pensioner accounts come from one outlet (IOL). A Joburg ETC article of the same date relays it and adds nothing independent.","IOL page dates were taken from the URLs (23 April and 24 May 2026) because the page bodies show only relative ages."],"geography":{"basis":"IOL reports social grant recipients across South Africa struggling with the portal, and the SASSA statement and parliamentary reply concern the South African Social Security Agency's beneficiaries.","court_countries":[],"event_countries":["ZA"],"affected_person_countries":["ZA"]},"publication":{"basis":"IOL (23 April and 24 May 2026), SASSA's 10 April 2026 statement on gov.za, Daily Voice and a Parliament committee statement were read in full. SASSA's and IOL's accounts are attributed and left at reported, and the pensioners' failures come from one outlet. The pensioners are not named. No inspected report attributes any grant suspension to facial verification failures.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"The sources describe a facial recognition step in SASSA's eLife Certification portal, which uses biometric verification through the eKYC system and integration with Home Affairs systems, and IOL's 24 May 2026 report of a parliamentary reply refers to SASSA's electronic facial biometric system without saying whether it is the eLife portal. The vendor and model are not identified. The department attributes facial verification problems to poor lighting, unstable connectivity or missing Home Affairs biometric records, and a SASSA spokesperson says the portals work with intermittent downtime, so the cause of any individual failure is not established.","status":"reported"},"person_relations":["made_decision_about"]},"name":"South Africa: pensioners report repeated failures of the facial recognition step in SASSA's eLife certification portal, and SASSA reports disruptions and office queues","summary":"The South African Social Security Agency (SASSA) introduced an online eLife Certification (life certification) for grant beneficiaries that uses biometric verification through its electronic Know Your Client (eKYC) system. IOL reports the certification was implemented on 30 March 2026. SASSA says beneficiaries who do not complete life certification as directed may face payment delays or suspension. On 10 April 2026 SASSA apologised to beneficiaries who could not access the portal, said system glitches linked to interfaces with other departments had caused delays, disruptions and long queues at its offices, and said the problem was resolved. On 23 April 2026 IOL reported that a pensioner couple said they had tried the facial recognition option 22 times since 2 April without success, and that beneficiaries nationwide told IOL they could not complete the certification, citing failures with facial recognition and one-time PINs, with one pensioner also reporting a message that Home Affairs was not available to verify their particulars. A SASSA spokesperson said the portals work and that 13,644 (88%) of the 15,499 unique clients who accessed the online verification services by 16 April were verified, and IOL reports SASSA admitted the system has been working intermittently. In a May 2026 report on a parliamentary reply, IOL said SASSA stated that unsuccessful facial recognition attempts on online platforms were among the causes of non-verification (those beneficiaries are redirected to fingerprint checks at local offices) and that it had recorded 7,779 complaints linked to its electronic facial biometric system. The department attributed facial verification issues to poor lighting, unstable connectivity or missing biometric records at Home Affairs. Neither May report mentions the eLife portal, and IOL places the figures within a biometric verification rollout that it dates from September 2025. The reports do not say how many grants were suspended because of facial verification failures.","incidentDate":"2026-03-30","incidentKind":"bounded_series","incidentDatePrecision":"day","exposurePattern":"repeated_interactions","reportedDate":"2026-04-23","aiSystem":"Facial recognition step in SASSA's eLife Certification portal (biometric verification through the eKYC system, integrated with Home Affairs systems). Vendor and model not identified in the sources.","aiProduct":"SASSA eLife facial verification","aiCompany":"South African Social Security Agency (SASSA), deployer","severity":"low","verificationStatus":"unverified","harmCategories":[],"harmOutcomes":["other_material_harm"],"harmOutcomeSummary":"SASSA said glitches caused delays, disruptions and long queues at its offices, and a pensioner couple told IOL they had tried the facial recognition step 22 times since 2 April without success. SASSA warned that beneficiaries who do not complete life certification as directed may face payment delays or suspension. No inspected report says that either pensioner's grant was delayed or suspended.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":2,"affectedCountStatus":"partial","affectedCountEvidence":"IOL reports one pensioner couple (two people) who said the facial recognition step failed 22 times. IOL says it was inundated with emails and calls from beneficiaries nationwide and SASSA recorded 7,779 complaints, but complaints are not people and the number of beneficiaries who failed facial verification is not reported, so no larger count is recorded. The 67,868 grant suspensions are not counted because no inspected report attributes them to facial verification.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://iol.co.za/news/south-africa/2026-04-23-sassa-elife-certification-portal-beneficiaries-report-ongoing-glitches-despite-agency-denials/","primarySourceLabel":"IOL (23 Apr 2026)","firstPublishedAt":"2026-09-29T21:16:13.146866+00:00","updatedAt":"2026-09-30T01:17:54.03839+00:00","scopeVersion":"facts-v3","tags":["historical-2026"]},{"id":"2025-us-manhattan-airbnb-damage-claim-photos-disputed-as-manipulated-guest-refunded","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"the host told Airbnb that she had caused more than £12,000 worth of damage, and submitted pictures of an apparently cracked coffee table as part of his case.","relation":"supports","source_id":"s1"},{"locator":"Airbnb initially told her “after careful review of the photos”, she would have to reimburse the host a total of £5,314. She appealed against the decision.","relation":"supports","source_id":"s1"}],"assertion":"The host told Airbnb the guest had caused more than 12,000 pounds of damage, submitted pictures including an apparently cracked coffee table, and Airbnb initially told the guest after reviewing the photos that she must reimburse 5,314 pounds.","causal_attribution":"The Guardian relays the guest's account of the host's claim and Airbnb's first decision. Airbnb's ruling and the photos were not inspected. New York Post and Fox Business rewrite this account."},{"id":"c2","status":"reported","evidence":[{"locator":"A close examination of two pictures of the coffee table appears to show differences in the damage, leading the woman to believe they were digitally manipulated or generated by AI.","relation":"supports","source_id":"s1"},{"locator":"He did not respond to a request for comment.","relation":"supports","source_id":"s1"}],"assertion":"The Guardian reports that a close examination of two pictures of the coffee table appears to show differences in the damage, leading the guest to believe they were digitally manipulated or generated by AI. The host did not respond to the Guardian.","causal_attribution":"Guest's belief and the Guardian's description of her comparison. The photos were not inspected and no forensic finding was reported. Whether an AI tool was used is not established."},{"id":"c3","status":"reported","evidence":[{"locator":"Airbnb said he had been warned for violating its terms and told he would be removed if there was another similar report. The company told him that it could not verify the images he submitted as part of his complaint.","relation":"supports","source_id":"s1"},{"locator":"We have taken action with the host, and we are reviewing the original handling of this case.","relation":"supports","source_id":"s3"}],"assertion":"Airbnb told the host it could not verify the images he submitted, warned him for violating its terms and said he would be removed if there was another similar report.","causal_attribution":"Airbnb statements to the Guardian and to Fox Business. Airbnb did not state a finding that the images were AI-generated. The company's review of the photos is described only as 'careful review' with no mention of AI tools."},{"id":"c4","status":"reported","evidence":[{"locator":"Five days after Guardian Money raised questions about the case with Airbnb, the woman was told it had accepted her appeal and credited her account with £500.","relation":"supports","source_id":"s1"},{"locator":"the company offered an £854 refund – a fifth of the cost of her booking. She refused to accept this","relation":"supports","source_id":"s1"},{"locator":"was refunded the full cost (£4,269) of her booking, and a negative review that the host had placed on her profile was taken down.","relation":"supports","source_id":"s1"},{"locator":"Airbnb apologised and said there would be a review into how her case was handled.","relation":"supports","source_id":"s1"}],"assertion":"After Guardian Money asked Airbnb about the case, Airbnb accepted the appeal, credited 500 pounds, then refunded the full booking cost of 4,269 pounds after the guest refused a partial offer, the host's negative review on her profile was taken down (the Guardian does not say by whom), and Airbnb apologised.","causal_attribution":"The Guardian relays the guest's account and Airbnb's statements. The refund and review removal were not documented by Airbnb records seen here."},{"id":"c5","status":"reported","evidence":[{"locator":"experience fell below our usual high standards.","relation":"supports","source_id":"s3"},{"locator":"assure them that they will not be charged for the reported damage","relation":"supports","source_id":"s3"}],"assertion":"Airbnb said the guest's experience fell below its usual standards and that she would not be charged for the reported damage.","causal_attribution":"Airbnb's own statement to Fox Business."}],"effects":[{"label":"Guest told to reimburse the host on the strength of photos she says were manipulated; reversed on appeal","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.theguardian.com/technology/2025/aug/02/airbnb-guest-damage-claim-refund-photos","kind":"news","access":"read","language":"en","translation_note":"","independence_group":"guardian-report"},{"id":"s2","url":"https://nypost.com/2025/08/06/lifestyle/airbnb-guest-accused-of-causing-16k-in-property-damage-claims-host-used-ai-to-create-fake-photos/","kind":"news","access":"read","language":"en","translation_note":"","independence_group":"guardian-report"},{"id":"s3","url":"https://www.foxbusiness.com/lifestyle/airbnb-apologizes-superhost-allegedly-ai-doctored-photos-claim-16k-fake-damages","kind":"news","access":"read","language":"en","translation_note":"","independence_group":"airbnb-statement-to-fox"}],"version":1,"ai_roles":["others_use"],"contexts":["everyday_life","finance"],"unknowns":["Whether the host's photos were altered, and by what tool, is unknown: no forensic finding, host statement or Airbnb finding on AI is reported.","Airbnb's own review of the photos is described only as a 'careful review'. No source says Airbnb used an AI system to assess them.","The stay dates, the day of the damage claim and the outcome date are not stated beyond 'earlier this year' and the Guardian's 2 August 2025 date.","Whether the guest paid any part of the reimbursement before the appeal succeeded is not stated.","The US outlets give the claim as about 16,000 dollars (Fox Business) and the Guardian as more than 12,000 pounds. The inspected sources do not itemise the claim."],"geography":{"basis":"The Guardian places the apartment in Manhattan, New York, and describes the guest as London-based. Airbnb's location and the Guardian's publication are not used.","court_countries":[],"event_countries":["US"],"affected_person_countries":["GB"]},"publication":{"basis":"The Guardian report and two rewrites were read in full. The host and guest are not named and the apartment address is not given. The AI element rests on the guest's belief and Airbnb's statement that it could not verify the images. Claims stay at reported status.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"The guest says differences between two photos of one table lead her to believe they were manipulated or AI-generated. Airbnb told the host it could not verify the images. The host did not respond, and no source shows an AI tool was used. Fox Business and the New York Post headlines state AI use more strongly than the Guardian body supports.","status":"suspected"},"person_relations":["made_claim_about"]},"name":"Manhattan Airbnb: guest says host's damage-claim photos were manipulated or AI-generated, Airbnb first ordered payment then refunded her","summary":"The Guardian reported on 2 August 2025 that a London-based academic who rented a Manhattan apartment through Airbnb in 2025 was accused by the host of more than 12,000 pounds of damage, supported by photos including a cracked coffee table. She says differences between two photos of the table show they were digitally manipulated or AI-generated. Airbnb first told her to reimburse 5,314 pounds after reviewing the photos, then, five days after Guardian Money asked about the case, accepted her appeal, credited 500 pounds and, after she refused an 854-pound offer, refunded her booking cost of 4,269 pounds, and the host's negative review of her was taken down. Airbnb apologised. Airbnb told the host it could not verify the images he submitted and warned him. The host did not respond to the Guardian. No source establishes that AI was used.","incidentDate":"2025-01-01","incidentKind":"single_event","incidentDatePrecision":"year","exposurePattern":"single_interaction","reportedDate":"2025-08-02","aiSystem":"None identified (guest suspects the host's damage-claim photos were digitally manipulated or AI-generated; no tool or forensic finding is reported)","aiProduct":"Unidentified image tool (suspected)","severity":"low","verificationStatus":"unverified","harmCategories":[],"harmOutcomes":["other_material_harm","reputational_harm"],"harmOutcomeSummary":"The Guardian reports Airbnb initially told the guest to reimburse the host 5,314 pounds on the strength of the host's photos and the host placed a negative review on her profile; Airbnb later refunded her booking cost in full and the review was taken down.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"exact","affectedCountEvidence":"One guest reported subjected to the damage claim. Other Airbnb customers who might face similar claims are not counted.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://www.theguardian.com/technology/2025/aug/02/airbnb-guest-damage-claim-refund-photos","primarySourceLabel":"The Guardian (2 Aug 2025)","firstPublishedAt":"2026-09-29T13:28:51.915494+00:00","updatedAt":"2026-09-30T01:17:19.206649+00:00","scopeVersion":"facts-v3","tags":["historical-2025"]},{"id":"2025-us-eightfold-ai-match-score-applicant-screening-class-action","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"about unsuspecting job applicants and to score them from 0 to 5 for potential employers based on","relation":"supports","source_id":"s1"},{"locator":"This results in a Match Score, which ranges from 0 through 5 in increments of","relation":"supports","source_id":"s1"},{"locator":"Employers then use these reports to sift through","relation":"supports","source_id":"s1"},{"locator":"But candidates who apply for jobs at companies that use those tools are not given notice and a chance to dispute errors","relation":"supports","source_id":"s2"}],"assertion":"The complaint alleges that Eightfold collects applicant information and scores applicants from 0 to 5 by predicted likelihood of success for employers, that employers use these reports to sift applications, and that Eightfold provides no notice, access or dispute rights.","causal_attribution":"These are the plaintiffs' allegations as stated in a filing. The filing establishes only its own contents."},{"id":"c2","status":"reported","evidence":[{"locator":"In or around July 2025,","relation":"supports","source_id":"s1"},{"locator":"applied online via Microsoft’s website for","relation":"supports","source_id":"s1"},{"locator":"Manager-Responsible AI position two days after she applied, without being invited to interview.","relation":"supports","source_id":"s1"},{"locator":"On or around December 16, 2025,","relation":"supports","source_id":"s1"},{"locator":"received an automated rejection for the Senior Technical Program","relation":"supports","source_id":"s1"}],"assertion":"The complaint says one plaintiff applied to Microsoft in or around July 2025 and again around 16 December 2025, and received an automated rejection two days after the first application without an interview.","causal_attribution":"The complaint attributes the screening to Eightfold's Evaluation Tools used by the employer. No employer record or score was inspected."},{"id":"c3","status":"reported","evidence":[{"locator":"On or around December 17,","relation":"supports","source_id":"s1"},{"locator":"applied online via Paypal’s website","relation":"supports","source_id":"s1"},{"locator":"was not asked to interview and was not given a job offer for any of the","relation":"supports","source_id":"s1"}],"assertion":"The complaint says the other plaintiff applied to PayPal on or around 17 December 2025 for two roles and was not asked to interview or given an offer for any position applied to.","causal_attribution":"The complaint alleges Eightfold's tools were applied to this application. Reuters reports the plaintiffs believe the tools played a role. The link to any specific rejection is not established."},{"id":"c4","status":"reported","evidence":[{"locator":"is being sued in California for allegedly compiling reports used to","relation":"supports","source_id":"s2"},{"locator":"On January 20, 2026, Plaintiffs filed a Complaint in the Superior Court of","relation":"supports","source_id":"s1"},{"locator":"sued in California state court on behalf of all U.S. job seekers who applied for jobs","relation":"supports","source_id":"s2"},{"locator":"Neither was hired, and both believe that Eightfold's tools played a role.","relation":"supports","source_id":"s2"},{"locator":"The proposed class action was filed by two women with STEM backgrounds who believe Eightfold’s AI tools hurt their careers by unfairly screening them out","relation":"supports","source_id":"s3"}],"assertion":"Reuters reports the plaintiffs, who are not named by this record, filed a proposed class action in California state court on 20 January 2026 on behalf of all US job seekers evaluated with the tools, and that neither plaintiff was hired and both believe Eightfold's tools played a role.","causal_attribution":"Reuters and the plaintiffs' counsel release both trace to the complaint and counsel. They are one reporting chain."},{"id":"c5","status":"reported","evidence":[{"locator":"We do not scrape social media and the like.","relation":"supports","source_id":"s2"},{"locator":"said the platform operates on data shared by candidates or provided by customers.","relation":"supports","source_id":"s2"}],"assertion":"An Eightfold spokesperson said the platform does not scrape social media and operates on data shared by candidates or provided by customers.","causal_attribution":"Company statement to Reuters. It disputes the complaint's data-collection allegation and does not address the scoring allegation."},{"id":"c6","status":"documented","evidence":[{"locator":"Document 1 Filed 03/02/26","relation":"supports","source_id":"s1"},{"locator":"Signed by Judge Yvonne Gonzalez Rogers on 7/24/2026.","relation":"supports","source_id":"s4"},{"locator":"NOTICE OF REMOVAL Eightfold AI, Inc.'s Notice of Removal from Contra Costa Superior Court.","relation":"supports","source_id":"s4"},{"locator":"MOTION to Dismiss filed by Eightfold AI Inc..","relation":"supports","source_id":"s4"},{"locator":"ORDER VACATING HEARING Re: 29 MOTION to Dismiss.","relation":"supports","source_id":"s4"}],"assertion":"Eightfold removed the case to the US District Court for the Northern District of California on 2 March 2026 and filed a motion to dismiss. The docket shows the hearing on that motion vacated on 24 July 2026 and no ruling in the entries inspected.","causal_attribution":"Procedural facts documented by the court docket. The docket does not establish the merits."}],"effects":[{"label":"Applicants allegedly scored and screened by an undisclosed AI ranking without statutory notice or dispute rights, and not interviewed or hired","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://storage.courtlistener.com/recap/gov.uscourts.cand.465146/gov.uscourts.cand.465146.1.0_2.pdf","kind":"court_filing","access":"read","language":"en","translation_note":"","independence_group":"eightfold-complaint"},{"id":"s2","url":"https://finance.yahoo.com/news/ai-company-eightfold-sued-helping-181520410.html","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"eightfold-complaint"},{"id":"s3","url":"https://www.outtengolden.com/newsroom/landmark-class-action-accuses-eightfold-ai-of-illegally-producing-hidden-credit-reports-on-job-applicants","kind":"organisation_statement","access":"read","language":"en","translation_note":"","independence_group":"eightfold-complaint"},{"id":"s4","url":"https://www.courtlistener.com/docket/72351430/kistler-v-eightfold-ai-inc/","kind":"court_docket","access":"read","language":"en","translation_note":"","independence_group":"court-docket"}],"version":1,"ai_roles":["institutional_use"],"contexts":["work","privacy"],"unknowns":["Whether Eightfold's tools produced a score for either plaintiff or influenced either employer's decision is not established. The complaint and reports rest on the plaintiffs' allegations and belief.","The start date of the alleged practice is not stated. The incident range covers only the plaintiffs' dated applications (July to December 2025). The complaint says one plaintiff has applied to Microsoft and other employers since 2023. Earlier applications are not described and are outside this record.","Whether the defendant's tools collected social media or other third-party data about the plaintiffs is disputed by Eightfold and unresolved.","The other plaintiff's location is not stated in the sources inspected.","The outcome of the pending motion to dismiss is unknown. Docket entry 46 (24 Sep 2026, a statement of recent decision) had no readable content.","Microsoft and PayPal are not defendants. Microsoft declined comment and no PayPal response was inspected."],"geography":{"basis":"The complaint states that one plaintiff lives in, applied from and was denied employment while living in Contra Costa County, California. The removed case is in the US District Court for the Northern District of California. The other plaintiff's location is not stated.","court_countries":["US"],"event_countries":["US"],"affected_person_countries":["US"]},"publication":{"basis":"The complaint was read from the copy attached to the defendant's notice of removal, with Reuters, the plaintiffs' counsel release and the court docket. The plaintiffs are described by role and not named. All allegations are attributed and remain at reported status. The AIAAIC archive entry had no description and is not cited.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"The complaint alleges Eightfold's Match Score model scored and ranked the plaintiffs' applications for the employers. Eightfold's statement disputes the alleged data collection. No document showing a score for either plaintiff was inspected.","status":"reported"},"person_relations":["made_claim_about"]},"name":"US: two applicants sue Eightfold AI over 0-to-5 applicant scores used in hiring","summary":"Two job applicants filed a proposed class action on 20 January 2026 in California state court (removed to federal court on 2 March 2026) against Eightfold AI Inc. The complaint alleges that Eightfold's hiring tools collect applicant data and produce a 0-to-5 'Match Score' that employers use to rank candidates, without the notice, access and dispute rights the Fair Credit Reporting Act and a California statute require. One plaintiff says she applied to Microsoft in or around July 2025 and again in December 2025 and received an automated rejection two days after the first application. The other says she applied to PayPal in December 2025. Both say they were not interviewed or hired and believe Eightfold's tools played a role. Eightfold says it does not scrape social media and operates on data candidates or customers provide. The allegations are unproven and Eightfold filed a motion to dismiss (no ruling appears in the docket entries inspected).","incidentDate":"2025-07-01","incidentEndDate":"2025-12-17","incidentKind":"bounded_series","incidentDatePrecision":"range","exposurePattern":"repeated_interactions","reportedDate":"2026-01-21","aiSystem":"Eightfold AI talent-evaluation tools including the 'Match Score' applicant ranking model","aiProduct":"Eightfold Match Score","aiCompany":"Eightfold AI Inc.","severity":"low","verificationStatus":"unverified","harmCategories":[],"harmOutcomes":["professional_harm"],"harmOutcomeSummary":"The complaint alleges that two applicants were scored by Eightfold's tools, not interviewed and not hired, and that they received none of the disclosures or dispute rights the statutes require. The plaintiffs believe the tools played a role in their rejections. Eightfold's involvement in any particular rejection is not established.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":2,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"Two plaintiffs each allege they were scored and not hired. The proposed class of all US applicants evaluated with the tools is not counted because no class has been certified and no other applicant's experience is reported.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://storage.courtlistener.com/recap/gov.uscourts.cand.465146/gov.uscourts.cand.465146.1.0_2.pdf","primarySourceLabel":"Notice of removal with the state-court complaint, plaintiffs v. Eightfold AI Inc. (N.D. Cal., filed 2 Mar 2026)","firstPublishedAt":"2026-09-29T13:28:41.995961+00:00","updatedAt":"2026-09-30T01:17:19.000724+00:00","scopeVersion":"facts-v3","tags":["historical-2025"]},{"id":"2025-lacey-insurance-discovery-relief-denied-ai-citations","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"ECF 119 p. 8 para 20: 'I have struck, and decline to'; 'consider, any of the supplemental briefs'; 'I decline to award any of the discovery relief'; p. 7 para 17: 'The initial, undisclosed use of AI products to'; 'generate the first draft of the brief was flat-out wrong.'","relation":"supports","source_id":"s1"}],"assertion":"The special master struck the claimant's supplemental briefs and denied the discovery relief she sought because her lawyers filed briefs built on unverified AI-generated legal research.","causal_attribution":"The order itself connects the struck briefs and denial of relief to counsel's undisclosed AI use, failure to cite-check and re-submission of the defective brief (paras 17-20)."},{"id":"c2","status":"documented","evidence":[{"locator":"ECF 119 p. 2 para 5: 'approximately nine of the 27 legal'; 'citations in the ten-page brief were incorrect in some way. At least two of the'; 'authorities cited do not exist at all.'","relation":"supports","source_id":"s1"}],"assertion":"The order records that about nine of the 27 citations in the ten-page brief were incorrect and that at least two cited authorities do not exist.","causal_attribution":"The special master's own after-the-fact review, as recorded in the order."},{"id":"c3","status":"documented","evidence":[{"locator":"ECF 119 p. 10 para 24: 'clearly not at fault for the AI debacle'; 'She will not, however, be financially'; para 25: 'I decline to order any sanction'; conclusion: 'aggregate amount of $31,100.'","relation":"supports","source_id":"s1"}],"assertion":"The order makes the two firms jointly and severally pay $31,100 in aggregate, states the client was not at fault and will not be financially responsible for the awards, and declines any sanction against individual lawyers.","causal_attribution":"Allocation of financial responsibility is directly recorded in the order."},{"id":"c4","status":"reported","evidence":[{"locator":"ECF 119 appendix, declaration p. 46 para 3: 'generative AI capabilities found in CoCounsel and Westlaw Precision and'; p. 47 para 4: 'It is unclear to me whether'; 'research tools I utilized returned some erroneous information.'","relation":"supports","source_id":"s1"}],"assertion":"One lawyer's sworn declaration says the outline he circulated relied in part on generative AI features in CoCounsel, Westlaw Precision and Google Gemini, and that he cannot tell whether particular errors arose from his transcription or from a tool.","causal_attribution":"The declarant's own statement; the order does not determine which tool produced each error."},{"id":"c5","status":"reported","evidence":[{"locator":"ECF 119 appendix, joint response section II p. 76: 'None of the K&L Gates attorneys who worked on the subject brief used any'; order p. 3 para 7: 'the attorneys at K&L Gates didn’t'","relation":"supports","source_id":"s1"}],"assertion":"The second firm states that none of its lawyers who worked on the brief used or had access to AI tools or knew that AI had been used until the special master's inquiries.","causal_attribution":"The firm's own submission; the order records that the special master had no reason to doubt the sworn statements."}],"effects":[{"label":"Loss of requested discovery relief after the claimant's briefs were struck","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.lawnext.com/wp-content/uploads/2025/05/C.D.-Cal.-24-cv-05205-dckt-000119_000-filed-2025-05-06.pdf","kind":"court_order","access":"read","language":"en","translation_note":"English; read on 2026-10-09 from the retained PDF text of the 77-page filing: order pp. 2-10, appendix inventory p. 11, originating lawyer's declaration pp. 46-47 and the firms' joint response section II p. 76. PDF page 1 yielded no text. Not every appended exhibit was read.","independence_group":"lacey-ecf119"}],"version":1,"ai_roles":["others_use"],"contexts":["justice"],"unknowns":["Which tool, if any, produced each erroneous citation or quotation; the declarant says he cannot distinguish his transcription errors from tool output.","Whether the requested discovery relief would have been granted had the briefs been accurate.","The later course of the insurance action and where the AI-assisted drafting took place.","Paragraph 1 of the order (PDF page 1) was not inspected because the page yielded no extractable text."],"geography":{"basis":"The order was issued by a special master in the US District Court for the Central District of California. The record does not state where the AI-assisted drafting took place or where the claimant lives, so event and affected-person countries are left unknown.","court_countries":["US"],"event_countries":[],"affected_person_countries":[]},"publication":{"basis":"A filed court order directly establishes that the claimant's briefs were struck and her discovery relief denied because of AI-generated false citations in her lawyers' filings, with the monetary award placed on the firms. The tool identities are attributed to the declarant and kept as reported.","reviewed_on":"2026-10-09"},"ai_involvement":{"basis":"The filed order records that a lawyer used AI tools to generate the research outline for the claimant's brief and that the resulting brief contained false citations and quotations (about nine of 27 wrong, at least two non-existent). The special master read those authorities, found they did not exist, and on that basis struck the briefs and denied the discovery relief, writing that the use of AI affirmatively misled him. The declarant names CoCounsel, Westlaw Precision and Google Gemini but cannot say which tool produced each error, so the per-tool contribution is unestablished.","status":"supported"},"person_relations":["informed_decision_about"]},"name":"Insurance claimant loses discovery relief after her lawyers file AI-generated false citations; special master's order sanctions the two firms $31,100","summary":"In an order dated 5 May 2025 and filed on 6 May in Lacey v. State Farm (C.D. Cal.), a special master struck the claimant's supplemental briefs and denied the discovery relief she sought after her lawyers filed briefs built on unverified AI-generated legal research: about nine of 27 citations were wrong and at least two cited authorities did not exist. One lawyer's sworn declaration says the outline he circulated relied on generative AI features in CoCounsel, Westlaw Precision and Google Gemini, and that he could not tell whether particular errors came from his transcription or from a tool; co-counsel at the second firm say they used no AI tools and did not know AI had been used. The order makes the two firms jointly pay $31,100, states the client was not at fault and will not bear the awards, and imposes no penalty on individual lawyers.","incidentDate":"2025-04-01","incidentEndDate":"2025-05-05","incidentKind":"bounded_series","incidentDatePrecision":"month","exposurePattern":"unknown","reportedDate":"2025-05-06","aiSystem":"CoCounsel, Westlaw Precision and Google Gemini, named in one lawyer's sworn declaration as the generative AI tools used for the research outline; the order itself refers to 'various AI tools' and does not establish which tool produced each error","aiProduct":"CoCounsel, Westlaw Precision and Gemini (reported)","severity":"medium","verificationStatus":"verified","harmCategories":[],"harmOutcomes":["legal_harm"],"harmOutcomeSummary":"The special master struck the claimant's supplemental briefs and denied the discovery relief she sought after her lawyers filed briefs built on unverified AI-generated citations; the $31,100 award falls on the firms, not on her.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"One claimant whose briefs were struck and whose discovery relief was denied (order paras 20 and 24). She did not use the AI tools herself, so participant users is 0. The lawyers and firms bear the monetary award but are not counted as harmed persons.","victimAgeRange":"unknown","jurisdiction":"US","platformType":"assistant","primarySourceUrl":"https://www.lawnext.com/wp-content/uploads/2025/05/C.D.-Cal.-24-cv-05205-dckt-000119_000-filed-2025-05-06.pdf","primarySourceLabel":"Lacey v. State Farm General Insurance Co., C.D. Cal. 2:24-cv-05205, ECF 119: special master's sanctions order dated 5 May 2025, filed 6 May 2025 (with appendix)","firstPublishedAt":"2026-09-29T10:29:01.113121+00:00","updatedAt":"2026-10-09T03:12:38.363121+00:00","scopeVersion":"facts-v3","tags":["historical-2025","ai-citations","court-sanction","legal-research","consequential-response"]},{"id":"2026-surat-customs-penalty-ai-hallucinated-citations-supreme-court","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'The present appeal arose out of an Order-in-Original dated 8 October 2025 passed by the Additional Commissioner of Customs, Surat, imposing a penalty of ₹425,27,99,100 on the appellant under Section 114, Customs Act, 1962, for misdeclaring a consignment of natural diamonds as lab-grown diamonds to pay a lower tariff. The challenge to the said penalty before the Gujarat High Court failed, as the appeal was dismissed vide order dated 20 January 2026.'","relation":"supports","source_id":"s2"},{"locator":"'The penalty was imposed for allegedly misdeclaring a consignment of natural diamonds as lab-grown diamonds to attract a lower tariff.'; 'the High Court dismissed his challenge on January 20, 2026, prompting him to approach the Supreme Court.'","relation":"supports","source_id":"s1"}],"assertion":"On 8 October 2025 the Additional Commissioner of Customs, Surat, imposed a penalty of ₹425,27,99,100 on a diamond trader under Section 114 of the Customs Act, 1962, for allegedly misdeclaring natural diamonds as lab-grown diamonds, and the Gujarat High Court dismissed his challenge on 20 January 2026.","causal_attribution":"Facts of the proceedings as recited in the Supreme Court judgment, reported by legal outlets; the judgment itself was not read."},{"id":"c2","status":"reported","evidence":[{"locator":"'Upon verifying such judgments and articles, it does appear that the second respondent has relied upon case laws that are either non-existent or have fake citations. A further investigation has also revealed that some of the case laws relied upon by the second respondent, which do exist, do not lay down the ratio deduced therefrom and it appears to be a hallucination of AI,' the Court observed.","relation":"supports","source_id":"s1"},{"locator":"'Upon verification, the court found that the second respondent had \"relied upon case laws that are either non-existent or have fake citations.\" It further found that some case laws which did exist did not lay down the legal propositions attributed to them, and that this \"appears to be a hallucination of AI\".'","relation":"supports","source_id":"s3"}],"assertion":"Legal reports quote the Supreme Court finding that the penalty order relied on nonexistent case law or fake citations, and on existing cases that did not support the attributed legal propositions, which it described as apparent AI hallucination.","causal_attribution":"Finding of the Supreme Court after its own verification of the cited material, as quoted by the reports. The court attributes the defects to AI in qualified terms ('appears to be'); the tool is not named."},{"id":"c3","status":"reported","evidence":[{"locator":"'The use of AI in this case, resulting in reliance placed by the second respondent on material which is dubious, proves fatal to the sustainability of the order imposing penalty on the appellant. Such circumstances compel us to interfere and remand the case,' the Court said.","relation":"supports","source_id":"s1"},{"locator":"'The proceedings were revived for fresh adjudication by an officer of the same rank other than the officer who had passed the original order.'; 'The Court left it to the wisdom of the appointing authority to initiate such action against the author of the Order-in-Original dated 8 October 2025 as may be warranted in the circumstances, in accordance with law.'","relation":"supports","source_id":"s2"},{"locator":"'The Supreme Court on Wednesday, September 2, 2026, set aside a customs penalty of over ₹425 crore'; 'A bench of Justice Dipankar Dutta and Justice Sheel Nagu did not examine the merits of the customs case at all.'","relation":"supports","source_id":"s3"},{"locator":"'Верховний суд Індії 2 вересня 2026 року розглянув справу щодо використання штучного інтелекту працівником митниці, який послався на неіснуючі судові справи та фальшиві цитати, накладаючи штраф у розмірі близько $45 млн' (the Supreme Court of India on 2 September 2026 considered a case of a customs officer's use of AI, who cited non-existent cases and fake quotations while imposing a fine of about $45 million)","relation":"context","source_id":"s4"}],"assertion":"Legal reports say that on 2 September 2026 the Supreme Court treated reliance on dubious AI material as fatal to the penalty order, set aside that order and the High Court's order without deciding the merits, and revived the proceedings for fresh adjudication by a different officer of the same rank. The reports also say the court left any action against the order's author to the appointing authority.","causal_attribution":"Holding of the Supreme Court as reported."}],"effects":[{"label":"a ₹425 crore customs penalty order resting on non-existent and misattributed case law that the Supreme Court found appeared to be AI hallucination stood against the trader for about eleven months and was upheld by the Gujarat High Court","claim_id":"c1","direction":"negative"},{"label":"the trader had to take the case to the Supreme Court, which voided the order but revived the proceedings for fresh adjudication","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.barandbench.com/news/litigation/supreme-court-scraps-425-crore-penalty-over-fake-citations-ai-hallucinations-in-customs-order","kind":"news_report","access":"read","language":"en","translation_note":"Full body (including JSON-LD article body) read by curl on 2026-09-29. Reports and quotes the Supreme Court judgment of 2 September 2026. Reinspected in the retained body on 2026-10-06; no direct judgment obtained in this review.","independence_group":"sc-judgment-gadiya"},{"id":"s2","url":"https://www.scconline.com/blog/post/2026/09/03/supreme-court-sets-aside-%E2%82%B9425-crore-customs-penalty-over-ai-generated-fake-citations/","kind":"news_report","access":"read","language":"en","translation_note":"SCC Times case note, read by curl on 2026-09-29; summarises the same judgment (2026 SCC OnLine SC 1741). Reinspected in the retained body on 2026-10-06; no direct judgment obtained in this review.","independence_group":"sc-judgment-gadiya"},{"id":"s3","url":"https://upstox.com/news/personal-finance/tax/how-a-diamond-trader-got-relief-from-a-425-crore-penalty-after-customs-used-ai-generated-citations/article-199828/","kind":"news_report","access":"read","language":"en","translation_note":"Upstox (4 September 2026), read by curl on 2026-09-29; based on the same judgment. Reinspected in the retained body on 2026-10-06; no direct judgment obtained in this review.","independence_group":"sc-judgment-gadiya"},{"id":"s4","url":"https://sud.ua/uk/news/abroad/371427-mytnyk-poviryv-shi-i-vypysav-shtraf-na-45-mln-verkhovnyi-sud-use-skasuvav","kind":"news_report","access":"read","language":"uk","translation_note":"Ukrainian-language report (6 September 2026) read through the r.jina.ai relay on 2026-09-29 (direct fetch 403); translated by the reviewer. Cited only for context; it converts the penalty to about US$45 million. Reinspected in the retained body on 2026-10-06; no direct judgment obtained in this review. The Ukrainian article body was read with AI language assistance; it explicitly credits Supreme Court Observer and supplies context only.","independence_group":"sc-judgment-gadiya"}],"version":1,"ai_roles":["institutional_use"],"contexts":["justice","public_services","finance"],"unknowns":["Which AI tool the customs officer used and how it was used in drafting the order.","Whether the trader paid any part of the penalty or incurred other financial consequences while the order stood, and the trader's litigation costs.","The outcome of the revived adjudication and whether any action was taken against the officer.","The trader's country and place of residence."],"geography":{"basis":"The order was passed by the Additional Commissioner of Customs, Surat, challenged in the Gujarat High Court and decided by the Supreme Court of India. The trader's country of residence is not stated in the read reports, so affected_person_countries is left empty.","court_countries":["IN"],"event_countries":["IN"],"affected_person_countries":[]},"publication":{"basis":"Legal reporting describes a concrete penalty imposed on a trader through an administrative decision that used apparently AI-generated false authorities. The reported Supreme Court ruling connects reliance on that material to the decision to set the order aside. The trader is unnamed and the underlying customs allegation remains undecided. All cited accounts derive from the same judgment, whose text has not been inspected. No payment of the penalty or quantified financial loss is asserted.","reviewed_on":"2026-10-06"},"ai_involvement":{"basis":"Bar & Bench, SCC Times and Upstox report that the Supreme Court verified the authorities used in the officer's penalty order, found nonexistent cases, false citations and misattributed propositions, and described them as apparent AI hallucination. The reports connect this material to the officer's order imposing the penalty, which the court set aside on that basis. This supports informed_decision_about: the human officer used the material in an adverse decision about the trader. The tool and how it was used remain unknown, and the judgment itself has not been inspected.","status":"supported"},"person_relations":["informed_decision_about"]},"name":"India's Supreme Court overturns customs penalty citing apparent AI hallucinations","summary":"On 8 October 2025 the Additional Commissioner of Customs in Surat imposed a penalty of ₹425,27,99,100 (about ₹425 crore) on a diamond trader under Section 114 of the Customs Act, 1962, for allegedly misdeclaring a consignment of natural diamonds as lab-grown diamonds to pay a lower tariff. The trader's challenge was dismissed by the Gujarat High Court on 20 January 2026. Before the Supreme Court the trader argued that judgments and articles cited in the penalty order had been generated using artificial intelligence. On 2 September 2026 a bench of Justices Dipankar Datta and Sheel Nagu, after checking the material, found that the order relied on case law that was non-existent or carried fake citations, and on real cases that did not lay down the propositions attributed to them, which the court said 'appears to be a hallucination of AI'. Holding that the use of AI and reliance on this dubious material was fatal to the order, the court set aside both the penalty order and the High Court's order without deciding the merits, sent the matter for fresh adjudication by a different officer of the same rank, and left any action against the order's author to the appointing authority. The AI tool used is not named in the read reports.","incidentDate":"2025-10-08","incidentEndDate":"2026-09-02","incidentKind":"bounded_series","incidentDatePrecision":"range","exposurePattern":"unknown","reportedDate":"2026-09-02","aiSystem":"Unidentified AI tool reportedly used in authorities cited in a customs penalty order; legal reports quote the Supreme Court describing nonexistent and misattributed case law as apparent AI hallucination","aiProduct":"Unidentified AI tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["legal_harm"],"harmOutcomeSummary":"A diamond trader was subjected for about eleven months to a ₹425 crore customs penalty order, and lost a High Court challenge to it, before India's Supreme Court voided the order because it relied on non-existent and misattributed case law that the court found appeared to be AI hallucination (Supreme Court judgment as reported by Bar & Bench, SCC Online and Upstox). The underlying proceedings were revived.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"exact","affectedCountEvidence":"One trader, the appellant, on whom the penalty was imposed. The trader did not use the AI; the customs officer did. No other person is reported harmed.","victimAgeRange":"adult","jurisdiction":"IN","platformType":"other","outcomeType":"internal_action","outcomeStatus":"ongoing","primarySourceUrl":"https://www.barandbench.com/news/litigation/supreme-court-scraps-425-crore-penalty-over-fake-citations-ai-hallucinations-in-customs-order","primarySourceLabel":"Bar & Bench, 2 September 2026: Supreme Court scraps ₹425 crore penalty over fake citations, AI hallucinations in customs order","firstPublishedAt":"2026-09-29T09:08:50.773251+00:00","updatedAt":"2026-10-06T04:48:47.973979+00:00","scopeVersion":"facts-v3","tags":["india","surat","customs","supreme-court","hallucinated-citations","institutional-use","administrative-decision","penalty"]}]}