{ "name": "SafeLegalAI Global Legal AI Incident Tracker", "license": "CC BY 4.0 — cite safelegalai.com", "count": 72, "incidents": [ { "id": "dineen-shibata-v-kotchka", "caseName": "Dineen/Shibata v. Kotchka, No. 1 CA-CV 25-0606 PB (Ariz. Ct. App.)", "date": "2026-07-15T00:00:00.000Z", "court": "Arizona Court of Appeals, Division One", "jurisdiction": "us-state", "conduct": "Self-represented appellant Jeffrey Kotchka's opening brief, challenging a probate ruling, cited hallucinated legal authorities and misrepresented case citations that he obtained through generative AI ('Gen-AI') and did nothing to verify. After the appellee identified the fabrications and after a show-cause hearing, Kotchka neither moved to amend his brief nor withdrew the appeal, continuing to maintain claims unsupported by any valid authority.", "outcome": "sanctions", "sources": [ { "label": "Dineen/Shibata v. Kotchka, No. 1 CA-CV 25-0606 PB (Ariz. Ct. App. 15 Jul 2026), opinion of the Court (Furuya, J.; appeal affirmed, sanctions imposed) — CourtListener opinion text (read)", "url": "https://www.courtlistener.com/opinion/10931696/dineenshibata-v-kotchka/" } ], "summary": "The Arizona Court of Appeals affirmed a probate ruling and sanctioned self-represented appellant Jeffrey Kotchka, whose opening brief cited hallucinated and misrepresented authorities produced by generative AI. Holding that self-represented litigants must abide by the same rules, the court awarded appellee Kim Dineen the attorneys' fees and costs caused by the fake citations under A.R.S. §12-349 and ARCAP 25.", "status": "verified" }, { "id": "tobosaru-v-romania", "caseName": "Tobosaru v Court of Law Craiova, Romania, [2026] EWHC 1720 (Admin)", "date": "2026-07-08T00:00:00.000Z", "court": "High Court of England and Wales (King's Bench Division, Administrative Court)", "jurisdiction": "uk", "conduct": "In extradition proceedings the Crown Prosecution Service filed grounds of opposition and submissions citing two non-existent authorities, Vlad v Romania and Jeziorski v Poland. The CPS accepted the citations were inaccurate and likely originated from the use of artificial intelligence with inadequate human verification.", "outcome": "other", "sources": [ { "label": "Tobosaru v Court of Law Craiova, Romania [2026] EWHC 1720 (Admin), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ewhc/admin/2026/1720" } ], "summary": "In the High Court (Administrative Court), the Crown Prosecution Service filed extradition submissions citing two non-existent authorities, Vlad v Romania and Jeziorski v Poland, that the CPS later accepted likely originated from AI. The error was caught before the hearing; the appeal was dismissed on its merits and the CPS reviewed 78 other cases.", "status": "verified" }, { "id": "pooja-ramesh-singh-v-jk-bank", "caseName": "Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd. & Anr., 2026 INSC 668", "date": "2026-07-02T00:00:00.000Z", "court": "Supreme Court of India", "jurisdiction": "india", "conduct": "The NCLT and NCLAT decided an insolvency guarantee matter relying on precedents that did not exist — fake, AI-hallucinated judgments, some carrying wrong citations and others wholly invented. The Supreme Court found the tribunals' reasoning rested partly on this fabricated authority.", "outcome": "other", "sources": [ { "label": "Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd., 2026 INSC 668 (Civil Appeal No. 11950 of 2025, 2 July 2026) — Indian Kanoon (full judgment)", "url": "https://indiankanoon.org/doc/113338666/" } ], "summary": "India's Supreme Court (Justices Narasimha and Aradhe) set aside NCLT and NCLAT orders that had relied on fake, AI-hallucinated precedents, holding a decision resting even partly on fabricated authority is 'no decision in law'. It adopted a zero-tolerance approach, called citing unverified AI cases advocate misconduct, and asked the Bar Council of India to frame rules.", "status": "verified", "article": "pooja-ramesh-singh-india-supreme-court-ai" }, { "id": "leiske-v-kidd", "caseName": "Leiske v. Kidd, C.A. No. 2025-0426-CDW (LWW) (Del. Ch.)", "date": "2026-07-01T00:00:00.000Z", "court": "Court of Chancery of Delaware", "jurisdiction": "us-state", "conduct": "In an advancement dispute, plaintiffs' counsel's answering brief (filed 22 Jan 2026) contained fictitious citations, fabricated quotations and hallucinated legal propositions produced by an unspecified generative-AI ('GenAI') tool used to revise the brief. Counsel admitted failing to verify the citations, attributing the oversight to a paralegal's review; the 'corrected' brief merely removed quotation marks without fixing the false descriptions, and counsel faulted opposing counsel for raising the errors without a meet-and-confer.", "outcome": "other", "sources": [ { "label": "Leiske v. Kidd, C.A. No. 2025-0426-CDW (LWW) (Del. Ch. 1 Jul 2026), letter opinion of Vice Chancellor Lori W. Will (Rule to Show Cause) — CourtListener opinion text (read)", "url": "https://www.courtlistener.com/opinion/10885202/kevin-leiske-v-robert-gregory-kidd/" } ], "summary": "In the Delaware Court of Chancery, plaintiffs' counsel filed an answering brief with fictitious citations, fabricated quotations and hallucinated propositions from a generative-AI tool, then only stripped quotation marks when 'correcting' it. Vice Chancellor Lori W. Will issued a Rule 11(c) rule to show cause against signatory Richard P. Rollo and his firm Richards, Layton & Finger; sanctions are pending.", "status": "verified" }, { "id": "lnu-v-blanche", "caseName": "Lnu v. Blanche, No. 24-4790 (9th Cir.)", "date": "2026-06-03T00:00:00.000Z", "court": "US Court of Appeals for the Ninth Circuit", "jurisdiction": "us-federal", "conduct": "Attorneys Mike Singh Sethi and William Rounds, representing an immigration petitioner, filed briefs containing multiple nonexistent cases, misattributed quotations and gross misrepresentations of real authorities that were the product of generative-AI hallucinations. In a 'Motion to Correct' they characterized the fabrications as innocent typographical errors and affirmatively denied that generative AI might have been the source — a repeated failure of candor the panel found aggravating.", "outcome": "suspension", "monetaryPenalty": 5000, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "State Bar of California", "disposition": "The Ninth Circuit ordered the Clerk to serve its disciplinary order on the State Bar of California and any other applicable licensing authorities for further proceedings as appropriate. No state-bar disposition is yet on record.", "date": "2026-06-03T00:00:00.000Z" }, "sources": [ { "label": "Lnu v. Blanche, No. 24-4790 (9th Cir. 3 Jun 2026), disciplinary order (Paez, Bea, Forrest, JJ.) — CourtListener opinion text (read)", "url": "https://www.courtlistener.com/opinion/10869591/lnu-v-blanche/" } ], "summary": "The Ninth Circuit disciplined attorneys Mike Singh Sethi and William Rounds for filing immigration briefs with nonexistent cases, misattributed quotations and gross misrepresentations from AI hallucinations, then denying AI was the source. The panel suspended both from practice before the court for six months, fined each $2,500, and served its order on the State Bar of California.", "status": "verified" }, { "id": "oba-v-reeves", "caseName": "State ex rel. Oklahoma Bar Association v. Reeves, 2026 OK 37", "date": "2026-05-27T00:00:00.000Z", "court": "Supreme Court of Oklahoma", "jurisdiction": "us-state", "aiTool": "ChatGPT", "conduct": "Attorney Matthew Brett Reeves used ChatGPT to generate five fabricated citations added to two motions in the federal case Johnson v. Dunn (N.D. Ala.), for which he was sanctioned. Reciprocal bar discipline followed in several states.", "outcome": "warning", "regulatoryOutcome": { "body": "State bar disciplinary authorities (Oklahoma, Texas, Tennessee, Alabama)", "disposition": "Public reprimand by the Supreme Court of Oklahoma (27 May 2026) under reciprocal discipline, following a Texas public reprimand (4 Nov 2025), a Tennessee public censure (15 Dec 2025), and an Alabama public reprimand (6 Mar 2026).", "date": "2026-05-27T00:00:00.000Z" }, "sources": [ { "label": "State ex rel. Oklahoma Bar Association v. Reeves, 2026 OK 37 (27 May 2026) — CourtListener", "url": "https://www.courtlistener.com/opinion/10864894/state-of-oklahoma-ex-rel-oba-v-reeves/" } ], "summary": "The Oklahoma Supreme Court publicly reprimanded attorney Matthew Reeves under reciprocal-discipline rules after he admitted using ChatGPT to fabricate five citations in federal filings in Johnson v. Dunn — conduct already reprimanded or censured by the bars of Texas, Tennessee, and Alabama. A rare multi-state disciplinary chain over one AI incident.", "status": "verified" }, { "id": "cork-v-smith", "caseName": "Cork & Anor v Smith, [2026] EWHC 1199 (Ch)", "date": "2026-05-22T00:00:00.000Z", "court": "High Court of England and Wales (Business and Property Courts, Insolvency and Companies List)", "jurisdiction": "uk", "conduct": "Pinsent Masons LLP presented, as a direct quotation in a 30 March 2026 letter to the court, fabricated text purporting to be Insolvency Rule 12.37(5) — a provision that does not exist. A junior associate solicitor had used an AI system that hallucinated the rule; a second letter compounded the error.", "outcome": "referral", "regulatoryOutcome": { "body": "Solicitors Regulation Authority", "disposition": "Pinsent Masons self-referred to the SRA and agreed to pay the clients' costs arising from the erroneous correspondence; ICC Judge Mullen's judgment stands as a public admonishment, with the SRA investigation to follow.", "date": "2026-05-22T00:00:00.000Z" }, "sources": [ { "label": "Cork & Anor v Smith [2026] EWHC 1199 (Ch), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ewhc/ch/2026/1199" } ], "summary": "In the High Court (Insolvency and Companies List), Pinsent Masons LLP quoted a fabricated version of Insolvency Rule 12.37(5) — text that does not exist — after a junior solicitor used an AI system that hallucinated the provision. The firm self-referred to the Solicitors Regulation Authority and agreed to pay the clients' costs of the erroneous correspondence.", "status": "verified", "article": "pinsent-masons-sra-self-referral" }, { "id": "gleason-v-marcus", "caseName": "Gleason v. Marcus, No. 2D2024-2688 (Fla. 2d DCA)", "date": "2026-05-13T00:00:00.000Z", "court": "Florida Second District Court of Appeal", "jurisdiction": "us-state", "conduct": "Self-represented appellant Christopher Gleason, challenging the dismissal of his election-related complaint against the Pinellas County Supervisor of Elections, filed appellate briefs that cited numerous cases that do not exist and frequently quoted case law for propositions appearing nowhere in the authorities cited — which the court attributed to his apparent use of artificial intelligence to prepare the briefs.", "outcome": "warning", "sources": [ { "label": "Gleason v. Marcus, No. 2D2024-2688 (Fla. 2d DCA 13 May 2026), per curiam (Kelly, Rothstein-Youakim, Labrit, JJ.) — CourtListener opinion text (read)", "url": "https://www.courtlistener.com/opinion/10857994/gleason-v-marcus-canvassing-board-of-pinellas-county/" } ], "summary": "Florida's Second District Court of Appeal affirmed the dismissal of Christopher Gleason's pro se election complaint and admonished him for apparent AI use: his briefs cited numerous non-existent cases and quoted authorities for propositions found nowhere in them. The court warned that future filings of that nature would not receive favorable treatment and reminded all litigants to verify their filings.", "status": "verified" }, { "id": "rafique-v-hmrc", "caseName": "Rafique v Commissioners for HMRC, [2026] UKFTT 673 (TC)", "date": "2026-05-06T00:00:00.000Z", "court": "First-tier Tribunal (Tax Chamber)", "jurisdiction": "uk", "conduct": "A self-represented appellant sent emails citing several authorities (including purported Perrin, Hanuman Commercial, Vardy Properties and Moulsdale Properties decisions) that did not exist. Judge Bailey found the text was AI-generated and had not been reviewed by the appellant, describing the cases as AI hallucinations.", "outcome": "dismissal", "sources": [ { "label": "Rafique v HMRC [2026] UKFTT 673 (TC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukftt/tc/2026/673" } ], "summary": "Before the First-tier Tax Tribunal, self-represented appellant Omar Rafique sent emails citing multiple non-existent authorities that Judge Bailey found were AI-generated and unreviewed. The tribunal refused to reinstate his struck-out appeal and declined to issue a contempt summons, warning him against relying on AI-hallucinated cases in any future proceedings.", "status": "verified" }, { "id": "tuyunuklu-v-akmal", "caseName": "Tuyunuklu v Akmal, [2026] UKUT 174 (LC)", "date": "2026-05-05T00:00:00.000Z", "court": "Upper Tribunal (Lands Chamber)", "jurisdiction": "uk", "conduct": "An unrepresented appellant relied on Rakusen v Jepson [2023] UKSC 9 for a proposition the Supreme Court never decided. From references to AI in the hearing transcript the judge inferred the appellant may have been misled by an AI hallucination; AI use was suspected but not confirmed.", "outcome": "dismissal", "sources": [ { "label": "Tuyunuklu v Akmal [2026] UKUT 174 (LC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukut/lc/2026/174" } ], "summary": "Before the Upper Tribunal (Lands Chamber), unrepresented appellant Hakan Tuyunuklu relied on Rakusen v Jepson for a proposition the Supreme Court never decided. Judge Elizabeth Cooke dismissed his rent-repayment-order appeal and, noting references to AI in the hearing transcript, said she suspected he had been misled by an AI hallucination.", "status": "verified" }, { "id": "ibach-v-stewart", "caseName": "Ibach v. Stewart, No. SC-2025-0106 (Ala.)", "date": "2026-04-24T00:00:00.000Z", "court": "Supreme Court of Alabama", "jurisdiction": "us-state", "conduct": "W. Perry Hall, appellate counsel for the plaintiffs (Ibach and Stewart) in a trust dispute, filed briefs the Court found were riddled with misquoted and non-existent legal authorities — fabricated cases and quotations attributed to real courts that do not exist. The Supreme Court of Alabama held that 'the improper use of AI in the plaintiffs' briefs was widespread and particularly egregious,' cataloguing dozens of fictitious citations across the opening and reply briefs.", "outcome": "sanctions", "monetaryPenalty": 17200, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "Alabama State Bar", "disposition": "The Supreme Court of Alabama referred Hall to the Alabama State Bar for potential discipline over possible violations of the Alabama Rules of Professional Conduct. No Bar disposition is yet on record.", "date": "2026-04-24T00:00:00.000Z" }, "sources": [ { "label": "Ibach v. Stewart, No. SC-2025-0106 (Ala. 24 Apr 2026), opinion (appeal dismissed; sanctions on W. Perry Hall) — CourtListener opinion text (read)", "url": "https://www.courtlistener.com/opinion/10847586/laurie-ibach-and-mark-stewart-v-bruce-stewart-individually-as-trustee-of/" } ], "summary": "The Supreme Court of Alabama dismissed an appeal after finding plaintiffs' counsel W. Perry Hall filed briefs whose 'widespread and particularly egregious' improper use of AI produced dozens of non-existent and misquoted authorities. It ordered Hall to pay $17,200 plus double appellate costs, barred him from further unsigned filings, and referred him to the Alabama State Bar.", "status": "verified" }, { "id": "sullivan-cromwell-prince-global", "caseName": "In re Prince Global Holdings Ltd. (Sullivan & Cromwell AI disclosure), Ch. 15, S.D.N.Y.", "date": "2026-04-18T00:00:00.000Z", "court": "US Bankruptcy Court, Southern District of New York", "jurisdiction": "us-federal", "conduct": "Sullivan & Cromwell disclosed to Chief Judge Martin Glenn that its 9 April 2026 emergency Chapter 15 motion contained fabricated case citations, misquoted authorities and non-existent sources — AI 'hallucinations' its review process failed to catch, flagged by opposing counsel. The firm filed a corrected motion with a redline.", "outcome": "other", "sources": [ { "label": "Reported by multiple legal outlets (Law360, Global Legal Post/FT, Canadian Lawyer); primary letter/docket is on SDNY Bankruptcy PACER — [NEED: PACER primary before status can be verified]", "url": "https://www.globallegalpost.com/news/sullivan-cromwell-apologises-for-ai-hallucinations-in-letter-to-us-court-1300253238" } ], "summary": "In the US Bankruptcy Court for the Southern District of New York, Sullivan & Cromwell self-reported to Chief Judge Martin Glenn in April 2026 that its Chapter 15 emergency motion in In re Prince Global Holdings contained AI-hallucinated citations and misquotes. Opposing counsel flagged them; the firm filed corrections. No sanction was imposed or reported pending.", "status": "unverified" }, { "id": "mccarthy-v-dea", "caseName": "McCarthy v. U.S. Drug Enforcement Administration (3d Cir. 2026), No. 24-2704", "date": "2026-03-27T00:00:00.000Z", "court": "US Court of Appeals for the Third Circuit", "jurisdiction": "us-federal", "aiTool": "AI (unspecified)", "conduct": "Petitioner's counsel filed an opening brief with incorrect, hallucinated case summaries and one nonexistent authority generated by AI, and failed to correct them after the government flagged the errors. The court reprimanded counsel under its Disciplinary Rules and directed notification of all courts of his admission and the National Disciplinary Data Bank.", "outcome": "warning", "sources": [ { "label": "Stephen McCarthy v. DEA (3d Cir., 27 Mar 2026), No. 24-2704 — CourtListener", "url": "https://www.courtlistener.com/opinion/10826367/stephen-mccarthy-v-dea/" } ], "summary": "The US Court of Appeals for the Third Circuit reprimanded an attorney whose opening brief contained AI-hallucinated case summaries and a nonexistent authority that he failed to correct. As its first AI-hallucination case, the court withheld monetary sanctions but warned that future violators may face them.", "status": "verified" }, { "id": "guerin-v-odoherty", "caseName": "Guerin v O'Doherty, [2026] IECA 48", "date": "2026-03-26T00:00:00.000Z", "court": "Court of Appeal (Ireland)", "jurisdiction": "ireland", "conduct": "A self-represented appellant used AI to prepare her written submissions, which cited authorities that did not exist — described by the court as hallucinations generated by the AI system — and did not tell the opposing solicitors or the court that she had used AI. Counsel wasted time trying to locate the fabricated cases.", "outcome": "dismissal", "sources": [ { "label": "Guerin v O'Doherty [2026] IECA 48, Courts Service of Ireland (courts.ie)", "url": "https://www2.courts.ie/view/judgments/da0ff352-10c8-43da-8f24-ff4058366f2d/d9395c85-7585-48dc-8e1f-f21765426a02/2026_IECA_48.pdf/pdf" } ], "summary": "In the Irish Court of Appeal, self-represented appellant Gemma O'Doherty filed AI-prepared submissions citing authorities that did not exist and never disclosed her use of AI. Costello P dismissed her appeal and issued Ireland's first appellate guidance on AI in litigation: five principles requiring responsible use, disclosure of AI use, and independent verification of every authority.", "status": "verified" }, { "id": "ben-cohen-v-ramat-gan", "caseName": "Ben Cohen v. Municipality of Ramat Gan & Ministry of Education, AAM 63194-08-25", "date": "2026-03-22T00:00:00.000Z", "court": "Supreme Court of Israel (sitting as the Court of Administrative Appeals)", "jurisdiction": "israel", "conduct": "In an administrative dispute over school transport for a child with special needs, the Ramat Gan municipality relied on AI-generated materials that included a non-existent Ministry of Education director-general circular and fabricated court rulings. The court found the cited sources did not exist and had been fabricated by a generative-AI system, describing the reliance as reckless.", "outcome": "costs-order", "monetaryPenalty": 30000, "penaltyCurrency": "ILS", "sources": [ { "label": "Ben Cohen v. Municipality of Ramat Gan, AAM 63194-08-25 (Israeli Supreme Court, 22 Mar 2026) — official court decisions database (Hebrew; decision confirmed on server, Hebrew text corroborated via multiple expert readings — [pin-cite pending Hebrew-capable check])", "url": "https://supremedecisions.court.gov.il/Home/Download?path=NetVerdicts/2026/3/22/2025-8-63194-8-2&fileName=ca9f017678224cff8112b3cc1b864595&type=4" } ], "summary": "In Israel, the Supreme Court (sitting in administrative appeals, 22 March 2026) held that the Municipality of Ramat Gan had recklessly relied on generative-AI output — a non-existent Education Ministry directive and fabricated rulings — in a case over a special-needs child's school transport, and exceptionally ordered 30,000 NIS in costs, extending AI-hallucination accountability to a public body.", "status": "verified" }, { "id": "prososki-v-regan", "caseName": "Prososki v. Regan, 321 Neb. 38 (2026)", "date": "2026-03-20T00:00:00.000Z", "court": "Nebraska Supreme Court", "jurisdiction": "us-state", "conduct": "In a marital-dissolution appeal, appellant's counsel filed a brief in which 57 of 63 citations were defective — including roughly 20 AI 'hallucinations', three entirely fabricated cases, and fabricated quotations — asserted by the appellee to have been generated by generative AI.", "outcome": "referral", "regulatoryOutcome": { "body": "Nebraska Counsel for Discipline", "disposition": "Appellant's counsel W. Gregory Lake was referred for investigation of potential violations of the Rules of Professional Conduct (competence, diligence, candor toward the tribunal, and dishonesty). [NEED: primary-source confirmation of the reported April 2026 indefinite suspension of Mr. Lake — on the follow-up watch list.]", "date": "2026-03-20T00:00:00.000Z" }, "sources": [ { "label": "Prososki v. Regan, 321 Neb. 38 (20 Mar 2026), Nebraska Judicial Branch (official case page)", "url": "https://nebraskajudicial.gov/courts/supreme-court/supreme-court-call/prososki-v-regan" } ], "summary": "The Nebraska Supreme Court struck an appellant's brief in a divorce appeal after finding 57 of 63 citations defective — including about 20 AI hallucinations and three fabricated cases — dismissed the appeal, and referred counsel W. Gregory Lake to the Counsel for Discipline, holding that fictitious authority is sanctionable under existing candor and competence rules whether or not AI produced it.", "status": "verified" }, { "id": "state-v-coleman", "caseName": "State v. Coleman, 2026-Ohio-965 (Ohio Ct. App., 11th Dist.)", "date": "2026-03-20T00:00:00.000Z", "court": "Ohio Court of Appeals, Eleventh Appellate District", "jurisdiction": "us-state", "aiTool": "ChatGPT", "conduct": "In an application to reopen a criminal appeal, appellant's counsel filed a brief in which a staff member had used ChatGPT, generating hallucinated quotations not found in the cited authorities.", "outcome": "sanctions", "monetaryPenalty": 2000, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "Ohio Office of Disciplinary Counsel", "disposition": "The court fined counsel $2,000, removed him as counsel and struck the reopening application, and referred his conduct to the Office of Disciplinary Counsel of the Supreme Court of Ohio.", "date": "2026-03-20T00:00:00.000Z" }, "sources": [ { "label": "State v. Coleman, 2026-Ohio-965 (Ohio Ct. App., 11th Dist., 20 Mar 2026) — CourtListener", "url": "https://www.courtlistener.com/opinion/10812807/state-v-coleman/" } ], "summary": "An Ohio appellate court fined a criminal-appeal attorney $2,000, removed him as counsel, struck his reopening application, and referred him to the Ohio Office of Disciplinary Counsel after his staff's ChatGPT use produced hallucinated quotations. The opinion defends the court's authority to sanction independent of formal bar proceedings.", "status": "verified" }, { "id": "von-geitz-v-kelly", "caseName": "Von Geitz v Kelly & Ors; Von Geitz v Robertson & Ors, [2026] IECA 29", "date": "2026-03-16T00:00:00.000Z", "court": "Court of Appeal (Ireland)", "jurisdiction": "ireland", "conduct": "A self-represented plaintiff's written submissions across two appeals were, per Allen J, 'littered with' unsupported propositions, wrong citations 'and a few non-existent cases'. At the appeal hearing it was suggested the non-existent cases were hallucinations generated by AI; the court stressed that verifying them remained the litigant's responsibility.", "outcome": "dismissal", "sources": [ { "label": "Von Geitz v Kelly & Ors and Von Geitz v Robertson & Ors [2026] IECA 29 (Court of Appeal, Allen J., 16 Mar 2026; Record 2025/140 & 2025/178), Courts Service of Ireland — approved judgment PDF", "url": "https://www.courts.ie/view/judgments/494fb506-aaee-49eb-b7dd-da00b8e56364/c4916450-7a66-4eb1-9d4f-477da28d8ccf/2026_IECA_29.pdf/pdf" } ], "summary": "In the Irish Court of Appeal, Allen J found self-represented plaintiff Juerg Von Geitz's submissions across two appeals contained unsupported propositions, wrong citations and 'a few non-existent cases' suggested to be AI hallucinations. He dismissed both appeals, affirmed the High Court's strike-out and Isaac Wunder orders, and stressed the litigant's own duty to verify AI output.", "status": "verified" }, { "id": "green-v-imprint-creative", "caseName": "Green v Imprint Creative Print Solutions Ltd, 1809293/2024", "date": "2026-03-09T00:00:00.000Z", "court": "Employment Tribunal (Leeds)", "jurisdiction": "uk", "conduct": "A self-represented claimant, responding by email to a time-limit point, cited case law that neither the tribunal nor the respondent's counsel could identify. She readily accepted she had used AI to generate the submissions; the tribunal found the authorities did not exist.", "outcome": "dismissal", "sources": [ { "label": "Ms F Green v Imprint Creative Print Solutions Ltd: 1809293/2024, GOV.UK Employment Tribunal Decisions", "url": "https://www.gov.uk/employment-tribunal-decisions/ms-f-green-v-imprint-creative-print-solutions-ltd-1809293-slash-2024" } ], "summary": "Before the Leeds Employment Tribunal, self-represented claimant Ms F Green emailed submissions citing case law that neither the tribunal nor the respondent's counsel could identify. She accepted she had used AI to generate them; Employment Judge Armstrong found the authorities did not exist, disregarded them, and dismissed her disability and unfair-dismissal complaints.", "status": "verified" }, { "id": "tan-hai-peng-v-tan-cheong-joo", "caseName": "Tan Hai Peng Micheal and another v Tan Cheong Joo and another [2026] SGHC 49", "date": "2026-03-06T00:00:00.000Z", "court": "General Division of the High Court of the Republic of Singapore", "jurisdiction": "singapore", "aiTool": "ChatGPT", "conduct": "Court submissions contained fictitious AI-generated authorities. The court held that citing fictitious authorities is sanctionable whether or not AI produced them: the counsel on record had a non-delegable duty to verify, and the supervising solicitor who reviewed the draft also failed. The workload and supervision defence was rejected, each practitioner held individually accountable.", "outcome": "costs-order", "monetaryPenalty": 10000, "penaltyCurrency": "SGD", "sources": [ { "label": "Tan Hai Peng Micheal v Tan Cheong Joo [2026] SGHC 49 (6 Mar 2026) — Singapore Judiciary (elitigation.sg)", "url": "https://www.elitigation.sg/gd/s/2026_SGHC_49" } ], "summary": "In this Singapore High Court (General Division) costs ruling, S Mohan J made personal costs orders of S$5,000 each against the counsel on record and his supervising solicitor for filing fictitious AI-generated citations — S$10,000 within a global S$180,000 award. The judgment is a landmark on supervision-chain responsibility for generative-AI misuse.", "status": "verified" }, { "id": "ferko-v-beyond-reach", "caseName": "Ferko v Beyond Reach Ltd t/a Car Wash Crew, ADJ-00060622", "date": "2026-03-02T00:00:00.000Z", "court": "Workplace Relations Commission", "jurisdiction": "ireland", "conduct": "A litigant-in-person complainant's WRC complaint form was prepared by his sister using AI. He accepted he had not fully reviewed or verified the contents before submission and could not clearly explain aspects of the complaint advanced in his own name at the hearing.", "outcome": "dismissal", "sources": [ { "label": "ADJ-00060622, Workplace Relations Commission (workplacerelations.ie)", "url": "https://www.workplacerelations.ie/en/cases/2026/march/adj-00060622.html" } ], "summary": "Before Ireland's Workplace Relations Commission, complainant Imrich Ferko's complaint form was prepared by his sister using AI, and he admitted not verifying it before submission. Adjudication Officer Christina Ryan gave his evidence materially reduced weight, preferred the respondent, and held the complaint not well-founded, while noting that assisted complaint preparation is not itself improper.", "status": "verified" }, { "id": "jones-v-family-court-whangarei", "caseName": "Jones v Family Court at Whangārei [2026] NZSC 1", "date": "2026-02-11T00:00:00.000Z", "court": "Supreme Court of New Zealand", "jurisdiction": "nz", "conduct": "A self-represented litigant's leave-application submissions cited multiple hallucinated authorities — real case names paired with wrong citations, plus genuine cases misattributed — produced by an AI application. The Supreme Court granted an extension of time but dismissed the leave application and used the judgment to warn about AI misuse in litigation.", "outcome": "dismissal", "sources": [ { "label": "Jones v Family Court at Whangārei [2026] NZSC 1 (SC167/2025) — Courts of New Zealand (official judgment PDF)", "url": "https://www.courtsofnz.govt.nz/assets/cases/2026/2026-NZSC-1.pdf" } ], "summary": "New Zealand's Supreme Court dismissed a self-represented litigant's leave application after his submissions cited hallucinated authorities from an AI application — real case names with wrong citations and misattributed cases. The Court awarded NZ$500 costs and warned that reliance on unverified AI outputs may, in serious cases, amount to obstruction of justice or contempt of court.", "status": "verified" }, { "id": "psa-v-nmc-tchampet", "caseName": "Professional Standards Authority v Nursing and Midwifery Council & Anor, [2026] EWHC 141 (Admin)", "date": "2026-01-30T00:00:00.000Z", "court": "High Court of England and Wales (King's Bench Division, Administrative Court)", "jurisdiction": "uk", "aiTool": "Microsoft Copilot", "conduct": "An unregulated representative acting for the second respondent filed written argument partly generated by Microsoft Copilot, citing three authorities (Fatnani, Gupta and Ghosh) whose purported quotations were fabricated and which did not support his propositions. He admitted the references were phantoms created by AI.", "outcome": "warning", "sources": [ { "label": "Professional Standards Authority v NMC [2026] EWHC 141 (Admin), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ewhc/admin/2026/141" } ], "summary": "In the High Court (Administrative Court), unregulated representative Simon Holborn cited three authorities partly generated by Microsoft Copilot whose quotations were fabricated. Griffiths J warned him of contempt but, as a struck-off solicitor, he could not be referred to a regulator. The Authority's appeal succeeded and the nurse's suspension became a striking-off.", "status": "verified" }, { "id": "folarin-v-immigration-services-commissioner", "caseName": "Folarin v Immigration Services Commissioner, [2026] UKFTT 135 (GRC)", "date": "2026-01-29T00:00:00.000Z", "court": "First-tier Tribunal (General Regulatory Chamber)", "jurisdiction": "uk", "aiTool": "ChatGPT", "conduct": "A self-represented immigration adviser appealing the loss of his registration cited multiple authorities the tribunal could not locate in any legal database. He disclosed asking ChatGPT to identify the top cases supporting his arguments, then edited the summaries without reading the underlying judgments.", "outcome": "dismissal", "sources": [ { "label": "Folarin v Immigration Services Commissioner [2026] UKFTT 135 (GRC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukftt/grc/2026/135" } ], "summary": "Before the First-tier Tribunal (General Regulatory Chamber), immigration adviser Raphael Folarin, appealing the loss of his registration, cited multiple non-existent authorities obtained by asking ChatGPT for supporting cases and editing summaries he never verified. The tribunal dismissed his appeal, finding the fabricated citations confirmed his unfitness, and noted such conduct may amount to contempt.", "status": "verified" }, { "id": "cassata-v-macrina", "caseName": "Cassata v. Michael Macrina Architect, P.C., 2026 NY Slip Op 26014", "date": "2026-01-27T00:00:00.000Z", "court": "New York Supreme Court, Suffolk County", "jurisdiction": "us-state", "aiTool": "generative AI (AI-enhanced Westlaw)", "conduct": "Defense counsel filed an opposition containing nonexistent cases, fake quotations, and unsupported propositions largely plagiarized from another firm's AI-generated brief; the supervising partner and firm failed to supervise. Counsel denied using AI and the court disbelieved her.", "outcome": "sanctions", "monetaryPenalty": 10000, "penaltyCurrency": "USD", "sources": [ { "label": "Cassata v. Michael Macrina Architect, P.C., 2026 NY Slip Op 26014 (Sup. Ct. Suffolk Cnty., 27 Jan 2026) — CourtListener", "url": "https://www.courtlistener.com/opinion/10782930/cassata-v-michael-macrina-architect-pc/" } ], "summary": "A New York trial court fined an associate and her supervising partner $1,000 each (to the Lawyers' Fund for Client Protection) and ordered the firm to pay $8,000 in fees after an opposition brief plagiarized from an AI-generated filing containing nonexistent cases and quotations. The court struck the opposition but declined to refer counsel to bar discipline.", "status": "verified" }, { "id": "huish-v-hmrc", "caseName": "Huish v Commissioners for HMRC, [2026] UKFTT 129 (TC)", "date": "2026-01-16T00:00:00.000Z", "court": "First-tier Tribunal (Tax Chamber)", "jurisdiction": "uk", "conduct": "A self-represented appellant cited four purported authorities (Collins, Harrison, Baxter and Hicks) said to have been decided in 2021; none existed. He confirmed to the tribunal that he had used AI to help prepare his hearing materials.", "outcome": "other", "sources": [ { "label": "Huish v HMRC [2026] UKFTT 129 (TC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukftt/tc/2026/129" } ], "summary": "Before the First-tier Tax Tribunal, self-represented appellant Robert Huish cited four non-existent authorities he had generated using AI while preparing his hearing materials. The tribunal attached no blame to him given his litigant-in-person status but recorded the fake case names to warn others, and allowed his appeal on its substantive merits.", "status": "verified" }, { "id": "deutsche-bank-v-letennier", "caseName": "Deutsche Bank Natl. Trust Co. v. LeTennier, 2026 NY Slip Op 00040 (App. Div. 3d Dept.)", "date": "2026-01-08T00:00:00.000Z", "court": "New York Supreme Court, Appellate Division, Third Department", "jurisdiction": "us-state", "aiTool": "generative AI (unspecified)", "conduct": "Defense counsel filed appellate briefs, motions, and letters containing no fewer than 23 fabricated cases across five filings, plus numerous misrepresentations of real cases, and continued to add fake authorities after being alerted.", "outcome": "sanctions", "monetaryPenalty": 10000, "penaltyCurrency": "USD", "sources": [ { "label": "Deutsche Bank Natl. Trust Co. v. LeTennier, 2026 NY Slip Op 00040 (App. Div. 3d Dept., 8 Jan 2026) — CourtListener", "url": "https://www.courtlistener.com/opinion/10771750/deutsche-bank-natl-trust-co-v-letennier/" } ], "summary": "In what it called the first New York appellate decision addressing generative-AI sanctions, the Third Department fined defense counsel Joshua Douglass $7,500 and the defendant $2,500 under 22 NYCRR 130-1.1 after briefs contained 23 fabricated cases and counsel kept filing fake authorities after being alerted — a state intermediate-appellate precedent-setter.", "status": "verified" }, { "id": "elden-v-hmrc", "caseName": "Elden v Commissioners for HMRC, [2026] UKFTT 41 (TC)", "date": "2026-01-08T00:00:00.000Z", "court": "First-tier Tribunal (Tax Chamber)", "jurisdiction": "uk", "conduct": "The appellant's representatives — Strauss Phillips & Co, chartered accountants regulated by the ICAEW — filed a skeleton argument, prepared with AI, that materially misstated three real authorities (Hok, Atlantic Electronics and Leeds City Council v HMRC). The representative later confirmed AI had been used.", "outcome": "other", "sources": [ { "label": "Elden v HMRC [2026] UKFTT 41 (TC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukftt/tc/2026/41" } ], "summary": "Before the First-tier Tax Tribunal, Gary Elden's representatives, chartered accountants regulated by the ICAEW, filed a skeleton argument that used AI and misstated three real authorities. The tribunal refused HMRC's strike-out application but imposed strict directions requiring full judgment texts, pinpoint citations and statements of truth identifying who verified each cited case.", "status": "verified" }, { "id": "couvrette-v-wisnovsky", "caseName": "Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL (D. Or.)", "date": "2025-12-12T00:00:00.000Z", "court": "US District Court, District of Oregon (Medford Division)", "jurisdiction": "us-federal", "conduct": "Lead counsel Stephen Brigandi filed three summary-judgment briefs containing fifteen non-existent cases and fabricated quotations falsely attributed to eight real authorities, generated by generative AI. When defendants flagged the fakes, the amended briefs deceptively deleted only the quotation marks rather than correcting the fabrications. Magistrate Judge Mark D. Clarke imposed terminating sanctions; the court denied Plaintiffs' motion for reconsideration on 22 June 2026, leaving the sanctions in place.", "outcome": "sanctions", "monetaryPenalty": 110204, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "Oregon State Bar", "disposition": "The court directed the Clerk to transmit its 12 December 2025 sanctions order to the Oregon State Bar (Brigandi appeared pro hac vice and is a California attorney). Referral only — no Bar disposition is yet on record.", "date": "2025-12-12T00:00:00.000Z" }, "sources": [ { "label": "Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL (D. Or.), Opinion & Order (ECF 212, 12 Dec 2025) and fees Opinion & Order (ECF 225, 23 Mar 2026) — order PDFs read via CourtListener/RECAP", "url": "https://www.courtlistener.com/docket/55220104/couvrette-v-wisnovsky/" } ], "summary": "In the US District Court for Oregon, lead counsel Stephen Brigandi filed three briefs with fifteen non-existent cases and quotations falsely attributed to eight real authorities, then concealed them. Magistrate Judge Mark D. Clarke dismissed the claims, imposed $15,500 in sanctions plus $94,704 in fees (~$110,000 total, shared with local counsel), and referred Brigandi to the Oregon State Bar.", "status": "verified", "article": "couvrette-v-wisnovsky" }, { "id": "ko-v-li", "caseName": "Ko v. Li, 2025 ONSC 2965; 2025 ONSC 6785", "date": "2025-12-04T00:00:00.000Z", "court": "Ontario Superior Court of Justice", "jurisdiction": "canada", "aiTool": "ChatGPT", "conduct": "Applicant's counsel Jisuh Lee filed a factum citing fabricated cases hallucinated by ChatGPT and relied on them orally. An initial contempt was treated as purged after she blamed staff (2025 ONSC 2965, 20 May 2025); Lee then admitted, in a 30 Sept 2025 letter, that she had drafted the factum herself and had lied about it out of fear and embarrassment — a fresh deception of the court (2025 ONSC 6785, 4 Dec 2025).", "outcome": "referral", "regulatoryOutcome": { "body": "Law Society of Ontario", "disposition": "The Law Society of Ontario is investigating Lee's conduct (file CAS-165781-ZOC5S2) and attended the December 2025 case conference as an observer; no disposition yet. Separately, Justice Myers issued a fresh show-cause for criminal contempt and referred carriage to the Attorney General of Ontario; the Crown took carriage and an amicus curiae (Dean Embry) was appointed. No finding of contempt yet.", "date": "2025-12-04T00:00:00.000Z" }, "sources": [ { "label": "Ko v. Li, 2025 ONSC 2965 (CanLII)", "url": "https://www.canlii.org/en/on/onsc/doc/2025/2025onsc2965/2025onsc2965.html" }, { "label": "Ko v. Li, 2025 ONSC 6785 (CanLII)", "url": "https://www.canlii.org/en/on/onsc/doc/2025/2025onsc6785/2025onsc6785.html" } ], "summary": "In Ontario, lawyer Jisuh Lee filed a ChatGPT-fabricated factum in Ko v. Li and initially avoided contempt on conditions (2025 ONSC 2965). After she admitted lying about authorship, Justice Myers reopened the matter (2025 ONSC 6785), referred a criminal-contempt prosecution to Ontario's Attorney General, and the Law Society of Ontario opened an investigation.", "status": "verified", "article": "ko-v-li" }, { "id": "uk-v-sshd-2026-ukut-81", "caseName": "UK v Secretary of State for the Home Department (AI hallucinations; supervision; Hamid), [2026] UKUT 81 (IAC)", "date": "2025-11-17T00:00:00.000Z", "court": "Upper Tribunal (Immigration and Asylum Chamber)", "jurisdiction": "uk", "aiTool": "Google AI", "conduct": "Two immigration matters featured fabricated authorities traced to AI, including Google's AI results. In one, grounds drafted by a caseworker cited four non-existent cases; in the other an adviser cited a fictitious authority. The tribunal stressed the risk is not confined to generative models such as ChatGPT.", "outcome": "referral", "regulatoryOutcome": { "body": "Solicitors Regulation Authority", "disposition": "Solicitor and COLP Zubair Rasheed was referred to the SRA for failing to supervise the caseworker who drafted the fake citations. A separate adviser, Tahir Mohammed, escaped referral by self-reporting to the Immigration Advice Authority and the SRA before the tribunal acted.", "date": "2025-11-17T00:00:00.000Z" }, "sources": [ { "label": "UK v SSHD [2026] UKUT 81 (IAC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukut/iac/2026/81" } ], "summary": "Before the Upper Tribunal (Immigration and Asylum Chamber), two immigration matters featured fabricated authorities traced to AI, including Google's AI results. Solicitor and COLP Zubair Rasheed was referred to the Solicitors Regulation Authority for failing to supervise the caseworker who drafted fake citations; a separate adviser escaped referral by self-reporting to his regulators.", "status": "verified" }, { "id": "sheppard-v-jillion-qfc", "caseName": "Sheppard v Jillion LLC [2025] QIC (E) 3 (CTFIC0033/2025)", "date": "2025-11-12T00:00:00.000Z", "court": "Qatar Financial Centre Civil and Commercial Court (First Instance Circuit)", "jurisdiction": "gulf", "conduct": "A Dubai-based lawyer, acting on an extension-of-time application, cited two non-existent QFC authorities. He first claimed he could not access the judgments online, then admitted the citations came from Google search screenshots. President Lord Thomas of Cwmgiedd found the conduct intentional, not inadvertent, and a contempt of court under Article 35.3.", "outcome": "other", "sources": [ { "label": "Jonathan David Sheppard v Jillion LLC [2025] QIC (E) 3 (12 Nov 2025) — QFC Civil and Commercial Court (official judgment)", "url": "https://www.qicdrc.gov.qa/judgments/ctfic00332025" } ], "summary": "In this Qatar Financial Centre Civil and Commercial Court judgment, President Lord Thomas found an anonymised Dubai-based lawyer in contempt for citing two fake QFC cases sourced from Google and persisting after challenge, breaching Article 35.2. The court imposed no penal sanction, treating the apology and publication of the judgment as sufficient.", "status": "verified" }, { "id": "mezu-v-mezu", "caseName": "Mezu v. Mezu (Md. App. Ct. 2025), No. 361 Sept. Term 2025", "date": "2025-10-29T00:00:00.000Z", "court": "Appellate Court of Maryland", "jurisdiction": "us-state", "aiTool": "ChatGPT", "conduct": "The mother's appellate brief cited multiple fictitious AI-generated cases and cases not supporting the propositions cited. At oral argument, counsel admitted he typically did not read the cases he cited.", "outcome": "referral", "regulatoryOutcome": { "body": "Maryland Attorney Grievance Commission", "disposition": "The court referred the attorney to the Maryland Attorney Grievance Commission based on the nature and severity of the conduct.", "date": "2025-10-29T00:00:00.000Z" }, "sources": [ { "label": "Mezu v. Mezu (Md. App. Ct., 29 Oct 2025) — CourtListener", "url": "https://www.courtlistener.com/opinion/10714052/mezu-v-mezu/" } ], "summary": "Maryland's intermediate appellate court referred an attorney to the state Attorney Grievance Commission after his client's brief cited multiple fictitious ChatGPT-generated cases and he conceded he routinely did not read the authorities he cited — a bar-discipline referral rather than a monetary sanction.", "status": "verified" }, { "id": "kmg-wires-v-nfac", "caseName": "KMG Wires Pvt Ltd v. National Faceless Assessment Centre, WP(L) 24366/2025 (2025:BHC-OS:19789-DB)", "date": "2025-10-06T00:00:00.000Z", "court": "High Court of Judicature at Bombay", "jurisdiction": "india", "aiTool": "generative AI (unspecified)", "conduct": "A faceless income-tax assessment order raising demand on about Rs 27.91 crore relied on non-existent, AI-generated case law and ignored the taxpayer's replies, breaching natural justice. The High Court held that quasi-judicial authorities must cross-verify AI output rather than rely on it blindly.", "outcome": "other", "sources": [ { "label": "KMG Wires Pvt Ltd v. National Faceless Assessment Centre (Bombay HC, 6 Oct 2025), 2025:BHC-OS:19789-DB — Indian Kanoon", "url": "https://indiankanoon.org/doc/181115899/" } ], "summary": "In India, the Bombay High Court (6 October 2025) quashed a faceless income-tax assessment raising demand on about Rs 27.91 crore because the assessing officer relied on non-existent, AI-generated precedents and ignored the taxpayer's evidence. The court warned quasi-judicial authorities not to blindly trust AI output and remanded the matter for fresh assessment.", "status": "verified" }, { "id": "in-re-ra-iowa", "caseName": "In re R.A. (Iowa Ct. App. 2025), No. 24-1629", "date": "2025-10-01T00:00:00.000Z", "court": "Court of Appeals of Iowa", "jurisdiction": "us-state", "aiTool": "AI-driven research tools (unspecified)", "conduct": "In a private child-in-need-of-assistance termination appeal, mother's counsel filed a reply brief citing nonexistent cases and statutes and inaccurate quotations after relying on secondary AI-driven research tools when she had Westlaw login problems.", "outcome": "referral", "monetaryPenalty": 150, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "Iowa Attorney Discipline Board", "disposition": "The court struck the reply brief, imposed a $150 penalty (or two hours of AI-ethics CLE in lieu), and directed the clerk to transmit the opinion and counsel's written explanation to the Iowa Attorney Discipline Board.", "date": "2025-10-01T00:00:00.000Z" }, "sources": [ { "label": "In the Interest of R.A. (Iowa Ct. App., 1 Oct 2025), No. 24-1629 — CourtListener", "url": "https://www.courtlistener.com/opinion/10685230/in-the-interest-of-ra-minor-child/" } ], "summary": "An Iowa appellate court struck a reply brief, imposed a $150 penalty (or an AI-ethics CLE alternative), and referred counsel to the Iowa Attorney Discipline Board after AI-driven research tools produced hallucinated citations. The court noted it had reviewed a database of more than 355 global AI-hallucination cases.", "status": "verified" }, { "id": "specter-aviation-v-laprade", "caseName": "Specter Aviation Limited c. Laprade, 2025 QCCS 3521", "date": "2025-10-01T00:00:00.000Z", "court": "Cour supérieure du Québec (Quebec Superior Court)", "jurisdiction": "canada", "aiTool": "ChatGPT (reported)", "conduct": "A self-represented litigant opposing homologation of an arbitral award is reported to have filed a contestation riddled with AI-generated non-existent case law and doctrine — eight instances of fabricated or irrelevant citations. The court reportedly found a serious breach of the conduct of proceedings under article 342 of the Code of Civil Procedure.", "outcome": "fine", "monetaryPenalty": 5000, "penaltyCurrency": "CAD", "sources": [ { "label": "Specter Aviation Limited c. Laprade, 2025 QCCS 3521 (CanLII, French) — primary not machine-readable (CanLII CAPTCHA); corroborated by Gowling WLG, McMillan, Norton Rose Fulbright, Lavery", "url": "https://www.canlii.org/fr/qc/qccs/doc/2025/2025qccs3521/2025qccs3521.html" } ], "summary": "In Quebec, a self-represented litigant in Specter Aviation c. Laprade is reported to have filed a contestation containing eight AI-generated fabricated or irrelevant citations produced with ChatGPT. The Quebec Superior Court reportedly imposed a CAD $5,000 penalty under article 342 of the Code of Civil Procedure, reportedly Quebec's first AI-misuse sanction. The primary judgment remains to be confirmed.", "status": "unverified" }, { "id": "reddy-v-saroya", "caseName": "Reddy v Saroya, 2025 ABCA 322", "date": "2025-09-26T00:00:00.000Z", "court": "Court of Appeal of Alberta", "jurisdiction": "canada", "aiTool": "generative AI (unnamed)", "conduct": "An appellant's factum contained seven non-existent authorities, six purportedly from the Alberta Court of Appeal, produced by a contractor the lawyer had hired who used generative AI. The Court held that the lawyer whose name appears on a filing bears ultimate responsibility for its contents, even where a contractor drafts it.", "outcome": "costs-order", "sources": [ { "label": "Reddy v Saroya, 2025 ABCA 322 (CanLII)", "url": "https://www.canlii.org/en/ab/abca/doc/2025/2025abca322/2025abca322.html" } ], "summary": "In Reddy v Saroya, the Court of Appeal of Alberta held that an appellant's factum containing seven fabricated authorities — produced by a contractor using generative AI — was the signing lawyer's responsibility. It signalled personal costs; a later order (reported as 2026 ABCA 20, CAD $17,550) is said to be Canada's first personal-costs order for AI hallucinations.", "status": "verified" }, { "id": "greenopolis-v-narender-singh", "caseName": "Greenopolis Welfare Association v. Narender Singh & Ors, CM(M) 1909/2025", "date": "2025-09-25T00:00:00.000Z", "court": "High Court of Delhi", "jurisdiction": "india", "conduct": "In a petition challenging trial-court interlocutory orders, the petitioner cited judicial precedents that do not exist and quoted passages absent from real judgments. Respondents' counsel flagged the citations as non-existent and AI-generated; the order records the fabrications but does not itself name a tool. (Press attributes them to ChatGPT.)", "outcome": "other", "sources": [ { "label": "Greenopolis Welfare Association v. Narender Singh & Ors, CM(M) 1909/2025 (Delhi HC, 25 Sep 2025) — Indian Kanoon", "url": "https://indiankanoon.org/doc/198343396/" } ], "summary": "In India, the Delhi High Court (25 September 2025) recorded that a petitioner's case relied on judicial precedents that do not exist and on quoted passages absent from real judgments — reported as ChatGPT-generated. Justice Kathpalia allowed the petition to be dismissed as withdrawn after respondents exposed the fabricated, AI-generated citations.", "status": "verified" }, { "id": "cingel-v-ferreri", "caseName": "Bethany Cingel v. Gregory Ferreri (Ind. Ct. App. 2025), No. 25A-DC-00500", "date": "2025-09-19T00:00:00.000Z", "court": "Court of Appeals of Indiana", "jurisdiction": "us-state", "aiTool": "generative AI (suspected)", "conduct": "A self-represented appellant's brief cited 23 authorities, 14 of them nonexistent. The court found she had likely used generative artificial intelligence, affirmed the decision below, and cautioned her but imposed no sanction.", "outcome": "warning", "sources": [ { "label": "Bethany Cingel v. Gregory Ferreri (Ind. Ct. App., 19 Sep 2025), No. 25A-DC-00500 — CourtListener", "url": "https://www.courtlistener.com/opinion/10674373/bethany-cingel-v-gregory-ferreri/" } ], "summary": "The Indiana Court of Appeals affirmed against a self-represented appellant in a domestic-relations appeal whose brief cited 23 authorities, 14 of them nonexistent, finding she had likely used generative AI. The court cautioned her about relying on unverified AI output but imposed no sanction, the opposing party having sought none.", "status": "verified" }, { "id": "noland-v-land-of-the-free", "caseName": "Noland v. Land of the Free, L.P., No. B331918 (Cal. Ct. App.)", "date": "2025-09-12T00:00:00.000Z", "court": "California Court of Appeal, Second Appellate District", "jurisdiction": "us-state", "aiTool": "generative AI (unspecified)", "conduct": "Nearly all legal quotations in plaintiff's opening brief, and many in the reply, were fabricated by generative-AI tools that counsel used and never verified. The court found the briefing riddled with citations to authority that did not say what was claimed, or did not exist.", "outcome": "sanctions", "monetaryPenalty": 10000, "penaltyCurrency": "USD", "regulatoryOutcome": { "body": "State Bar of California", "disposition": "The court directed the clerk of the court to serve a copy of its opinion on the State Bar of California, referring counsel for the fabricated AI-generated citations.", "date": "2025-09-12T00:00:00.000Z" }, "sources": [ { "label": "Noland v. Land of the Free, L.P., No. B331918 (Cal. Ct. App., 2d Dist., 12 Sep 2025) — California Courts (official opinion PDF)", "url": "https://www.courts.ca.gov/opinions/documents/B331918.PDF" } ], "summary": "In the California Court of Appeal, Second District, plaintiff's counsel filed opening and reply briefs in which nearly all quotations were fabricated by generative-AI tools counsel never verified. The court imposed a $10,000 monetary sanction and directed that its opinion be served on the State Bar of California.", "status": "verified" }, { "id": "williams-v-kirch", "caseName": "Kurt Williams v. Karin Kirch (Ind. Ct. App. 2025), No. 25A-SC-00196", "date": "2025-08-18T00:00:00.000Z", "court": "Court of Appeals of Indiana", "jurisdiction": "us-state", "aiTool": "generative AI (suspected)", "conduct": "A self-represented appellant cited fictitious authorities the court found were likely generated by AI. The opposing party sought no sanction, so the court admonished him without imposing a penalty.", "outcome": "warning", "sources": [ { "label": "Kurt Williams v. Karin Kirch (Ind. Ct. App., 18 Aug 2025), No. 25A-SC-00196 — CourtListener", "url": "https://www.courtlistener.com/opinion/10655761/kurt-williams-v-karin-kirch/" } ], "summary": "The Indiana Court of Appeals admonished a self-represented appellant who cited fictitious authorities the court found were likely produced by generative AI. Because the opposing party sought no sanction, the court imposed none, using the decision to warn litigants about relying on unverified AI output.", "status": "verified" }, { "id": "jne24-v-minister-immigration", "caseName": "JNE24 v Minister for Immigration and Citizenship [2025] FedCFamC2G 1314", "date": "2025-08-15T00:00:00.000Z", "court": "Federal Circuit and Family Court of Australia (Division 2), General Federal Law", "jurisdiction": "australia", "aiTool": "Claude (research) and Microsoft Copilot (validation)", "conduct": "In a migration judicial-review matter, a Western Australian lawyer filed submissions citing four cases that did not exist or did not support the propositions advanced. He admitted developing an overconfidence in relying on AI tools and assuming their output was reliable. The proceeding was discontinued and he reimbursed his client in full.", "outcome": "costs-order", "monetaryPenalty": 8371.3, "penaltyCurrency": "AUD", "regulatoryOutcome": { "body": "Legal Practice Board of Western Australia (LPBWA)", "disposition": "The court referred the lawyer's conduct to the Legal Practice Board of Western Australia; no LPBWA disposition has been published — the referral is pending. A personal costs order of AUD 8,371.30 was made against the lawyer, payable to the Minister.", "date": "2025-08-15T00:00:00.000Z" }, "sources": [ { "label": "JNE24 v Minister for Immigration and Citizenship [2025] FedCFamC2G 1314 (AustLII)", "url": "https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FedCFamC2G/2025/1314.html" } ], "summary": "In Australia's Federal Circuit and Family Court, a Western Australian lawyer filed migration submissions citing four fake authorities produced using Claude and 'validated' with Microsoft Copilot, admitting overconfidence in AI. Judge Gerrard ordered him to pay AUD 8,371.30 in personal costs and referred his conduct to the Legal Practice Board of Western Australia.", "status": "verified" }, { "id": "ms-bangladesh-2025-ukut-305", "caseName": "MS (Professional conduct, AI generated documents) Bangladesh, [2025] UKUT 305 (IAC)", "date": "2025-08-12T00:00:00.000Z", "court": "Upper Tribunal (Immigration and Asylum Chamber)", "jurisdiction": "uk", "aiTool": "ChatGPT", "conduct": "Counsel drafted grounds of appeal citing a fabricated Court of Appeal authority, Y (China) [2010] EWCA Civ 116, using ChatGPT and carried out no verification against reputable legal databases. He admitted using ChatGPT both when drafting the grounds and during the hearing.", "outcome": "referral", "regulatoryOutcome": { "body": "Bar Standards Board", "disposition": "The Upper Tribunal, presided over by its President Mr Justice Dove, referred barrister Muhammad Mujeebur Rahman to the Bar Standards Board for investigation, finding he had breached professional duties of honesty, integrity and competence.", "date": "2025-08-12T00:00:00.000Z" }, "sources": [ { "label": "MS (Bangladesh) [2025] UKUT 305 (IAC), HMCTS Tribunal Decisions (official UTIAC database)", "url": "https://tribunalsdecisions.service.gov.uk/utiac/2025-ukut-00305" } ], "summary": "Before the Upper Tribunal (Immigration and Asylum Chamber), barrister Muhammad Mujeebur Rahman cited a fabricated Court of Appeal case, Y (China), drafted using ChatGPT without verification. The tribunal, led by its President Mr Justice Dove, referred him to the Bar Standards Board for breaching duties of honesty, integrity and competence.", "status": "verified" }, { "id": "father-v-mother-2025-ewhc-2135-fam", "caseName": "The Father v The Mother & Ors, [2025] EWHC 2135 (Fam)", "date": "2025-07-30T00:00:00.000Z", "court": "High Court of England and Wales (Family Division)", "jurisdiction": "uk", "conduct": "A self-represented father applying for a judge's recusal submitted written authorities that the first-instance judge, HHJ Bailey, found were not genuine and appeared to have been generated by AI. An exact count of the fabricated cases is not specified in the judgment.", "outcome": "dismissal", "sources": [ { "label": "The Father v The Mother & Ors [2025] EWHC 2135 (Fam), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ewhc/fam/2025/2135" } ], "summary": "In the High Court (Family Division), a self-represented father seeking a judge's recusal submitted authorities that HHJ Bailey found were not genuine and appeared generated by AI. The case was referred to Mrs Justice Lieven as Family Presiding Judge, who dismissed the application, citing the Divisional Court's Ayinde ruling on fabricated citations.", "status": "verified" }, { "id": "hmrc-v-gunnarsson", "caseName": "HMRC v Gunnarsson, [2025] UKUT 247 (TCC)", "date": "2025-07-23T00:00:00.000Z", "court": "Upper Tribunal (Tax and Chancery Chamber)", "jurisdiction": "uk", "conduct": "An unrepresented respondent cited three non-existent First-tier Tribunal decisions (Patel, Ali and Kamran v HMRC) in his skeleton argument. He acknowledged using online AI software to prepare his submissions; the fabricated authorities were removed in an amended skeleton the same day.", "outcome": "dismissal", "sources": [ { "label": "HMRC v Gunnarsson [2025] UKUT 247 (TCC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukut/tcc/2025/247" } ], "summary": "Before the Upper Tribunal (Tax and Chancery Chamber), unrepresented respondent Marc Gunnarsson cited three non-existent First-tier Tribunal decisions produced by online AI software. The Upper Tribunal set aside the FTT decision, dismissed his appeal and confirmed recovery of £12,918 in wrongly claimed grants, warning that AI chatbots frequently mislead litigants in person.", "status": "verified" }, { "id": "dastou-v-holmes", "caseName": "Robert Dastou v. Kathryn Holmes, No. 2381CV02212 (Mass. Super. Ct.)", "date": "2025-07-16T00:00:00.000Z", "court": "Massachusetts Superior Court", "jurisdiction": "us-state", "aiTool": "ChatGPT", "conduct": "Defendant's counsel filed motions in limine and proposed jury instructions containing citations to nonexistent cases, a fabricated quotation, and a fictitious legal concept after using ChatGPT to 'format' her filings. The court held a hearing on the suspected misuse of generative AI.", "outcome": "sanctions", "sources": [ { "label": "Robert Dastou v. Kathryn Holmes (Mass. Super. Ct., 16 Jul 2025), Findings, Rulings and Order Imposing Sanctions — CourtListener", "url": "https://www.courtlistener.com/opinion/10642455/robert-dastou-v-kathryn-holmes/" } ], "summary": "A Massachusetts Superior Court sanctioned a defense attorney who used ChatGPT to 'format' filings, producing motions in limine and jury instructions riddled with hallucinated cases, a fake quotation, and a fictitious legal doctrine. The court accepted her agreement not to bill the client for the affected work and ordered her to complete a CLE course.", "status": "verified" }, { "id": "coomer-v-lindell", "caseName": "Coomer v. Lindell, No. 1:22-cv-01129 (D. Colo.)", "date": "2025-07-07T00:00:00.000Z", "court": "US District Court, District of Colorado", "jurisdiction": "us-federal", "conduct": "In the defamation suit brought by former Dominion director Eric Coomer, defense counsel filed a February 2025 opposition brief containing nearly thirty defective citations, including cases that did not exist, after running the draft through generative AI without checking it. Counsel did not readily disclose the AI use when the court asked.", "outcome": "sanctions", "monetaryPenalty": 6000, "penaltyCurrency": "USD", "sources": [ { "label": "Coomer v. Lindell, No. 1:22-cv-01129 (D. Colo.), sanctions order 7 July 2025 (Doc. 383) — CourtListener docket", "url": "https://www.courtlistener.com/docket/63296393/coomer-v-lindell/" } ], "summary": "In the US District Court for Colorado, lawyers for MyPillow CEO Mike Lindell filed an opposition brief with nearly thirty defective citations — including non-existent cases — produced by generative AI they did not verify. Judge Nina Y. Wang sanctioned Christopher Kachouroff and Jennifer DeMaster $3,000 each under Rule 11, and later added a $5,000 penalty for a repeat.", "status": "verified" }, { "id": "northbound-processing", "caseName": "Northbound Processing (Pty) Ltd v South African Diamond and Precious Metals Regulator (2025/072038) [2025] ZAGPJHC 661", "date": "2025-06-30T00:00:00.000Z", "court": "High Court of South Africa, Gauteng Division, Johannesburg", "jurisdiction": "south-africa", "aiTool": "Legal Genius", "conduct": "In heads of argument for an urgent licensing application, counsel cited non-existent authorities hallucinated by the AI tool 'Legal Genius', which claimed to be trained exclusively on South African law. Asked directly, counsel confirmed they appeared to be AI hallucinations. Smit AJ invoked the Article 16(1) judicial duty to report serious misconduct.", "outcome": "referral", "regulatoryOutcome": { "body": "Legal Practice Council (Gauteng Provincial Office)", "disposition": "The court referred the practitioners to the Legal Practice Council (Gauteng) for investigation, invoking the Article 16(1) reporting duty under the Code of Judicial Conduct; the LPC disposition is pending.", "date": "2025-06-30T00:00:00.000Z" }, "sources": [ { "label": "Northbound Processing (Pty) Ltd v SA Diamond and Precious Metals Regulator (2025/072038) [2025] ZAGPJHC 661 (30 Jun 2025) — SAFLII", "url": "https://www.saflii.org/za/cases/ZAGPJHC/2025/661.html" } ], "summary": "In this Gauteng High Court (South Africa) urgent mining-licence application, counsel's heads of argument cited fake authorities generated by the AI tool 'Legal Genius'. Smit AJ granted Northbound interim relief but, invoking the Article 16(1) reporting duty, referred the practitioners to the Legal Practice Council — the second such South African referral after Mavundla.", "status": "verified" }, { "id": "shahid-v-esaam", "caseName": "Nimat Shahid v. Sufyan Esaam (Ga. Ct. App. 2025)", "date": "2025-06-30T00:00:00.000Z", "court": "Court of Appeals of Georgia", "jurisdiction": "us-state", "aiTool": "generative AI", "conduct": "The husband's attorney, Diana Lynch, drafted a proposed trial-court order containing two fictitious AI-generated cases that the trial court adopted, then on appeal filed a brief citing roughly eleven more hallucinated or inapposite cases.", "outcome": "fine", "monetaryPenalty": 2500, "penaltyCurrency": "USD", "sources": [ { "label": "Nimat Shahid v. Sufyan Esaam (Ga. Ct. App., 30 Jun 2025), 918 S.E.2d 198 — CourtListener", "url": "https://www.courtlistener.com/opinion/10620062/nimat-shahid-v-sufyan-esaam/" } ], "summary": "The Georgia Court of Appeals penalized attorney Diana Lynch whose AI-hallucinated citations infected both a proposed trial-court order (adopted below) and her appellate brief — notable because fabricated cases were inserted into an actual court order. The court imposed a $2,500 frivolous-motion penalty, the statutory maximum.", "status": "verified" }, { "id": "ayinde-v-haringey", "caseName": "Ayinde v London Borough of Haringey; Al-Haroun v Qatar National Bank", "date": "2025-06-06T00:00:00.000Z", "court": "High Court of England and Wales (Divisional Court)", "jurisdiction": "uk", "conduct": "Submissions in two joined cases contained fabricated or misstated authorities apparently derived from generative AI; the Divisional Court used the Hamid jurisdiction to address the conduct of the lawyers involved.", "outcome": "referral", "regulatoryOutcome": { "body": "BSB / SRA", "disposition": "Barrister referred to the Bar Standards Board; solicitors referred to the Solicitors Regulation Authority. The court found the contempt threshold capable of being met but did not initiate proceedings, warning that leniency should not be treated as precedent.", "date": "2025-06-06T00:00:00.000Z" }, "sources": [ { "label": "Divisional Court judgment (judiciary.uk PDF, 6 June 2025)", "url": "https://www.judiciary.uk/wp-content/uploads/2025/06/Ayinde-v-London-Borough-of-Haringey-and-Al-Haroun-v-Qatar-National-Bank.pdf" } ], "summary": "The landmark England and Wales ruling on AI-fabricated citations: in joined cases, the Divisional Court invoked the Hamid jurisdiction over submissions containing non-existent authorities, referred a barrister to the BSB and solicitors to the SRA, and warned that the court's restraint on contempt should not be read as precedent.", "status": "verified", "article": "ayinde-v-haringey-al-haroun-divisional-court" }, { "id": "hussein-v-canada", "caseName": "Hussein v. Canada (Immigration, Refugees and Citizenship), 2025 FC 1060", "date": "2025-06-01T00:00:00.000Z", "court": "Federal Court of Canada", "jurisdiction": "canada", "aiTool": "Visto.ai (reported)", "conduct": "Applicant's counsel is reported to have relied on the immigration-law AI tool Visto.ai, which hallucinated two non-existent cases and mis-cited a real case for the governing legal test, and to have concealed the AI use. The court reportedly found a special reason for costs and signalled counsel may pay personally. [Exact date and details pending primary — CanLII CAPTCHA-blocked; absent from Federal Court Decisia.]", "outcome": "pending", "sources": [ { "label": "Hussein v. Canada (Citizenship and Immigration), 2025 FC 1060 (CanLII) — primary not machine-readable (CanLII CAPTCHA); corroborated by McCarthy Tétrault, Gowling WLG, Bennett Jones, Slaw", "url": "https://www.canlii.org/en/ca/fct/doc/2025/2025fc1060/2025fc1060.html" } ], "summary": "In Canada's Federal Court, applicant's counsel in an immigration matter is reported to have relied on the AI tool Visto.ai, which hallucinated the governing legal test and two non-existent cases, and to have concealed the AI use. The court reportedly found a special reason for costs and signalled personal liability; the primary judgment remains to be confirmed.", "status": "unverified" }, { "id": "r-v-chand", "caseName": "R. v. Chand, 2025 ONCJ 282", "date": "2025-05-26T00:00:00.000Z", "court": "Ontario Court of Justice", "jurisdiction": "canada", "conduct": "In a criminal trial, defence counsel is reported to have filed written final submissions containing a fictitious citation, case law that did not support the points cited, and unrelated civil cases — errors consistent with generative AI. The judge rejected the submissions and ordered counsel to prepare a new set without any generative AI.", "outcome": "other", "sources": [ { "label": "R. v. Chand, 2025 ONCJ 282 (CanLII) — primary not machine-readable (CanLII CAPTCHA); corroborated by CBC News, Law360 Canada, WeirFoulds", "url": "https://www.canlii.org/en/on/oncj/doc/2025/2025oncj282/2025oncj282.html" } ], "summary": "In Ontario, in the criminal trial R. v. Chand, defence counsel is reported to have filed final submissions containing a fictitious citation and unrelated authorities consistent with generative AI. Justice Kenkel rejected the submissions and directed counsel to prepare a fresh set with pinpoint citations and no generative-AI use. The primary judgment remains to be confirmed.", "status": "unverified" }, { "id": "garner-v-kadince", "caseName": "Garner v. Kadince, Inc., 2025 UT App 80", "date": "2025-05-22T00:00:00.000Z", "court": "Utah Court of Appeals", "jurisdiction": "us-state", "aiTool": "ChatGPT", "conduct": "Petitioner's counsel filed a petition containing fabricated authority generated by ChatGPT, including a non-existent 'Royer v. Nelson, 2007 UT App 74'. After an order to show cause, the court found the filing violated appellate rule 40. Counsel Richard Bednar acknowledged the fabricated authority came from ChatGPT.", "outcome": "sanctions", "monetaryPenalty": 1000, "penaltyCurrency": "USD", "sources": [ { "label": "Garner v. Kadince, 2025 UT App 80 (Utah Ct. App. 22 May 2025) — CourtListener (opinion PDF)", "url": "https://storage.courtlistener.com/pdf/2025/05/22/garner_v._kadince.pdf" } ], "summary": "In the Utah Court of Appeals, petitioner's counsel filed a brief containing ChatGPT-fabricated citations, including a non-existent 'Royer v. Nelson', violating appellate rule 40. The court sanctioned attorney Richard Bednar, ordering him to donate $1,000 to a legal-aid charity and to pay the respondents' fees and refund his client.", "status": "verified" }, { "id": "bandla-v-sra", "caseName": "Bandla v Solicitors Regulation Authority, [2025] EWHC 1167 (Admin)", "date": "2025-05-13T00:00:00.000Z", "court": "High Court of England and Wales (King's Bench Division, Administrative Court)", "jurisdiction": "uk", "conduct": "A former solicitor, appearing in person to appeal his removal from the roll, cited a large number of authorities in formal documents before the court; the SRA identified around 27 as non-existent. He denied using AI, attributing the citations to un-verified Google searches.", "outcome": "costs-order", "monetaryPenalty": 24727.2, "penaltyCurrency": "GBP", "regulatoryOutcome": { "body": "Solicitors Regulation Authority", "disposition": "The appellant had already been struck from the roll of solicitors by the SRA; the fabricated-authority conduct arose in his (unsuccessful) High Court appeal against that strike-off decision.", "date": "2025-05-13T00:00:00.000Z" }, "sources": [ { "label": "Bandla v SRA [2025] EWHC 1167 (Admin), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ewhc/admin/2025/1167" } ], "summary": "Before the High Court (Administrative Court), struck-off solicitor Venkateshwarlu Bandla cited around 27 non-existent authorities in his appeal against removal from the roll. Fordham J struck out his grounds as an abuse of process and ordered £24,727.20 in indemnity costs. Bandla denied using AI, attributing the fake citations to unverified Google searches.", "status": "verified" }, { "id": "an-v-archblock", "caseName": "An v. Archblock, Inc. (Del. Ch. 2025)", "date": "2025-04-04T00:00:00.000Z", "court": "Delaware Court of Chancery", "jurisdiction": "us-state", "aiTool": "generative AI (unspecified; court cited ChatGPT, Gemini, Copilot, CoCounsel as examples)", "conduct": "A self-represented petitioner's motion to compel cited false and miscited legal authority attributable to his use of generative AI. Vice Chancellor Lori Will denied the motion and, addressing the fabricated citations, entered a standing order for the litigation.", "outcome": "other", "sources": [ { "label": "Daniel Jaiyong An v. Archblock, Inc. (Del. Ch. 4 Apr 2025), Order — CourtListener (order PDF)", "url": "https://storage.courtlistener.com/pdf/2025/04/04/daniel_jaiyong_an_v._archblock_inc..pdf" } ], "summary": "In the Delaware Court of Chancery, a self-represented litigant's motion to compel relied on false citations produced by generative AI. Vice Chancellor Lori Will denied the motion and issued an order requiring litigants to certify their use of generative AI in future filings, warning that failure to comply may lead to sanctions.", "status": "verified" }, { "id": "dewald-ai-avatar", "caseName": "Dewald v. Massachusetts Mutual Life Ins. Co. (AI avatar at oral argument), App. Div. 1st Dept.", "date": "2025-03-26T00:00:00.000Z", "court": "New York Supreme Court, Appellate Division, First Department", "jurisdiction": "us-state", "aiTool": "AI-generated video avatar (tool not named)", "conduct": "A self-represented appellant obtained leave to play a prerecorded video for his oral argument, then played a generative-AI avatar of a fictional lawyer. The presiding justice stopped the video within seconds; the appellant admitted the presenter was AI-generated and was permitted to continue his argument in person.", "outcome": "warning", "sources": [ { "label": "Underlying appeal: Dewald v. Massachusetts Mutual Life Ins. Co., App. Div. 1st Dept. (Index No. 655380/23) — the avatar episode is oral-argument conduct, documented in press/argument record, not in the written decision", "url": "https://law.justia.com/cases/new-york/appellate-division-first-department/2025/index-no-655380-23-appeal-no-4138-case-no-2024-07337.html" } ], "summary": "Before New York's Appellate Division, First Department, self-represented appellant Jerome Dewald played an AI-generated avatar posing as his lawyer at his 26 March 2025 oral argument. The panel halted the video within seconds and he admitted it was AI-generated. He was verbally rebuked and allowed to continue in person; no sanction issued.", "status": "unverified" }, { "id": "sammaan-capital-v-mantri", "caseName": "Sammaan Capital Ltd v. Mantri Infrastructure Pvt Ltd, CRP No. 49 of 2025", "date": "2025-03-24T00:00:00.000Z", "court": "High Court of Karnataka, Bengaluru", "jurisdiction": "india", "conduct": "A Bengaluru City Civil Court judge, rejecting a jurisdictional objection, based the order on two Supreme Court decisions that were never delivered by any court. On revision, the High Court found the citations fabricated and directed action against the trial judge. No AI tool is named in the order.", "outcome": "referral", "sources": [ { "label": "Sammaan Capital Ltd v. Mantri Infrastructure Pvt Ltd, CRP No. 49/2025 (Karnataka HC, 24 Mar 2025) — Indian Kanoon", "url": "https://indiankanoon.org/doc/184681356/" } ], "summary": "In India, the Karnataka High Court (24 March 2025) found that a Bengaluru City Civil Court judge had decided a jurisdiction question by relying on two Supreme Court judgments that were never delivered. Justice Devdas called the conduct disturbing and directed that the order be placed before the Chief Justice for action against the trial judge.", "status": "verified" }, { "id": "wadsworth-v-walmart", "caseName": "Wadsworth v. Walmart, Inc., 348 F.R.D. 489 (D. Wyo. 2025)", "date": "2025-02-24T00:00:00.000Z", "court": "US District Court, District of Wyoming", "jurisdiction": "us-federal", "aiTool": "Morgan & Morgan internal AI tool (MX2.law)", "conduct": "Three plaintiffs' attorneys filed motions in limine on 22 January 2025 citing nine cases, eight of which did not exist. The fabricated authorities were produced using the firm's internal AI research tool; the drafter did not verify them before filing, and defendants noted some mis-cited cases also appear on ChatGPT.", "outcome": "sanctions", "monetaryPenalty": 5000, "penaltyCurrency": "USD", "sources": [ { "label": "Wadsworth v. Walmart Inc., No. 2:23-cv-00118 (D. Wyo.), sanctions order 24 Feb 2025 — CourtListener docket", "url": "https://www.courtlistener.com/docket/67564066/wadsworth-v-walmart-inc/" } ], "summary": "In the US District Court for Wyoming, three Morgan & Morgan attorneys filed motions in limine citing nine cases, eight non-existent, generated by the firm's internal AI tool. Judge Kelly Rankin fined lead drafter Rudwin Ayala $3,000 and revoked his pro hac vice status; T. Michael Morgan and Taly Goody were fined $1,000 each under Rule 11.", "status": "verified" }, { "id": "valu-v-minister-immigration", "caseName": "Valu v Minister for Immigration and Multicultural Affairs (No 2) [2025] FedCFamC2G 95", "date": "2025-01-31T00:00:00.000Z", "court": "Federal Circuit and Family Court of Australia (Division 2)", "jurisdiction": "australia", "aiTool": "ChatGPT", "conduct": "A solicitor filed an application and submissions citing Federal Court authorities that do not exist and fabricated quotes from an Administrative Appeals Tribunal decision, generated with ChatGPT and not verified. Judge Skaros found the conduct fell short of the standards of competence and diligence expected of a legal practitioner.", "outcome": "referral", "regulatoryOutcome": { "body": "NSW Office of the Legal Services Commissioner", "disposition": "The Principal Registrar was directed to refer the solicitor's conduct to the NSW Office of the Legal Services Commissioner.", "date": "2025-01-31T00:00:00.000Z" }, "sources": [ { "label": "Valu v Minister for Immigration and Multicultural Affairs (No 2) [2025] FedCFamC2G 95 (AustLII); (2025) 386 FLR 365", "url": "https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FedCFamC2G/2025/95.html" } ], "summary": "In one of Australia's first AI-citation cases, a solicitor before the Federal Circuit and Family Court cited non-existent Federal Court authorities and fabricated tribunal quotes generated by ChatGPT. Judge Skaros found the conduct fell short of competence and diligence standards and directed that it be referred to the NSW Office of the Legal Services Commissioner.", "status": "verified" }, { "id": "kohls-v-ellison", "caseName": "Kohls v. Ellison, No. 0:24-cv-03754 (D. Minn.)", "date": "2025-01-10T00:00:00.000Z", "court": "US District Court, District of Minnesota", "jurisdiction": "us-federal", "aiTool": "GPT-4o", "conduct": "In a First Amendment challenge to Minnesota's political-deepfakes statute, the State's expert — Stanford misinformation professor Jeff Hancock — filed a declaration drafted with GPT-4o that cited two non-existent academic articles and misattributed a third. Hancock admitted using AI and not verifying the citations.", "outcome": "other", "sources": [ { "label": "Kohls v. Ellison, No. 0:24-cv-03754 (D. Minn.), order 10 Jan 2025 (2025 WL 66514) — CourtListener docket", "url": "https://www.courtlistener.com/docket/69206960/kohls-v-ellison/" } ], "summary": "In the US District Court for Minnesota, an expert declaration defending the state's political-deepfakes law was itself tainted by AI: Stanford professor Jeff Hancock's GPT-4o-drafted report cited two fabricated academic articles and misattributed a third. Judge Laura Provinzino excluded the declaration, calling it ironic that an AI-misinformation expert fell for AI's 'siren call'.", "status": "verified" }, { "id": "mavundla-v-mec-kzn", "caseName": "Mavundla v MEC: Department of Co-Operative Government and Traditional Affairs, KZN (7940/2024P) [2025] ZAKZPHC 2", "date": "2025-01-08T00:00:00.000Z", "court": "High Court of South Africa, KwaZulu-Natal Division, Pietermaritzburg", "jurisdiction": "south-africa", "aiTool": "ChatGPT", "conduct": "On an application for leave to appeal, the applicant's supplementary notice cited multiple non-existent authorities sourced from ChatGPT. The research had been done by a candidate attorney and was not checked by the attorneys or counsel. Bezuidenhout J held that relying on AI for legal research without verification is irresponsible and unprofessional.", "outcome": "referral", "regulatoryOutcome": { "body": "Legal Practice Council (KwaZulu-Natal Provincial Office)", "disposition": "The court directed the registrar to send the judgment to the Legal Practice Council (KZN) for its attention and further action; the LPC's disposition is not yet on record. Wasted costs of two appearances were ordered against the firm Surendra Singh and Associates.", "date": "2025-01-08T00:00:00.000Z" }, "sources": [ { "label": "Mavundla v MEC, KZN (7940/2024P) [2025] ZAKZPHC 2 (8 Jan 2025) — SAFLII", "url": "https://www.saflii.org/za/cases/ZAKZPHC/2025/2.html" } ], "summary": "In this KwaZulu-Natal High Court (South Africa) leave-to-appeal ruling, the applicant's papers cited non-existent authorities generated via ChatGPT by a candidate attorney and unchecked by counsel. Bezuidenhout J called AI-reliance irresponsible, dismissed the application with costs, ordered wasted costs against the firm, and referred the matter to the Legal Practice Council.", "status": "verified" }, { "id": "buckeye-trust-itat", "caseName": "Buckeye Trust v. PCIT-2, Bangalore, ITA No. 1051/Bang/2024", "date": "2024-12-30T00:00:00.000Z", "court": "Income Tax Appellate Tribunal, Bengaluru ('A' Bench)", "jurisdiction": "india", "conduct": "The Tribunal's own order, resolving whether a private discretionary trust's receipt of investments worth about Rs 669 crore was taxable, relied in its reasoning on four case citations, three of which do not exist — two attributed to the Supreme Court and one to the Madras High Court. (Press attributes the fabrications to ChatGPT; the order does not name a tool.)", "outcome": "other", "sources": [ { "label": "Buckeye Trust v. PCIT-2, ITA No. 1051/Bang/2024 (ITAT Bengaluru, 30 Dec 2024) — Income Tax Appellate Tribunal (official order PDF)", "url": "https://itat.gov.in/public/files/upload/1735624480-urq7dV-1-TO.pdf" } ], "summary": "In India, the Bengaluru bench of the Income Tax Appellate Tribunal decided a roughly Rs 669 crore private-trust taxation appeal (Buckeye Trust, 30 December 2024) relying in its own reasoning on three non-existent case citations, two attributed to the Supreme Court and one to the Madras High Court. The Tribunal later recalled the order under section 254(2) for fresh hearing.", "status": "verified" }, { "id": "al-hamim-v-star-hearthstone", "caseName": "Al-Hamim v. Star Hearthstone, LLC, 2024 COA 128", "date": "2024-12-26T00:00:00.000Z", "court": "Colorado Court of Appeals", "jurisdiction": "us-state", "conduct": "A self-represented appellant filed a brief peppered with generative-AI-generated citations that did not exist. On a novel question for Colorado, the division declined to impose sanctions on these facts but issued a formal warning for the future.", "outcome": "warning", "sources": [ { "label": "Al-Hamim v. Star Hearthstone, LLC, 2024 COA 128 (Colo. App. 26 Dec 2024) — CourtListener (opinion PDF)", "url": "https://storage.courtlistener.com/pdf/2024/12/26/al-hamim_v._star_hearthstone_llc.pdf" } ], "summary": "In the Colorado Court of Appeals, a self-represented appellant's brief contained generative-AI hallucinated citations. On this novel question for Colorado, the division declined to impose sanctions on these facts but formally put lawyers and self-represented parties on notice that future filings containing AI-generated hallucinations may be sanctioned.", "status": "verified" }, { "id": "gauthier-v-goodyear", "caseName": "Gauthier v. Goodyear Tire & Rubber Co., No. 1:23-CV-281 (E.D. Tex.)", "date": "2024-11-25T00:00:00.000Z", "court": "US District Court, Eastern District of Texas (Beaumont Division)", "jurisdiction": "us-federal", "aiTool": "Claude (Anthropic)", "conduct": "Plaintiff's counsel Brandon Monk filed a summary-judgment response citing non-existent cases and fabricated quotations attributed to real cases. He admitted using the AI tool Claude to draft the brief and failing to verify the output, and initially took no corrective action after opposing counsel identified the fake authorities.", "outcome": "sanctions", "monetaryPenalty": 2000, "penaltyCurrency": "USD", "sources": [ { "label": "Gauthier v. Goodyear Tire & Rubber Co., No. 1:23-CV-281 (E.D. Tex. 25 Nov 2024), Order (ECF 48, Crone J.) — order PDF read via CourtListener/RECAP", "url": "https://www.courtlistener.com/docket/67624744/gauthier-v-goodyear-tire-rubber-co/" } ], "summary": "In the US District Court for the Eastern District of Texas, plaintiff's attorney Brandon Monk filed a brief with non-existent cases and fabricated quotations produced by the AI tool Claude, which he did not verify. Judge Marcia Crone ordered him to pay a $2,000 penalty into the court registry and complete a CLE course on generative AI.", "status": "verified" }, { "id": "brazil-trt2-ma-fe", "caseName": "Processo nº 1001467-35.2024.5.02.0467 (TRT-2, 4ª Turma)", "date": "2024-09-01T00:00:00.000Z", "court": "Tribunal Regional do Trabalho da 2ª Região (TRT-2), São Paulo (4ª Turma)", "jurisdiction": "brazil", "conduct": "A claimant's lawyer filed a labour appeal (September 2024) containing fabricated 'jurisprudence' — invented case summaries attributed to ministers of the Superior Labour Court (TST) and to a non-existent TRT-3 judge. The court confirmed none of the cited rulings existed; when questioned, counsel blamed the AI tool, which the rapporteur rejected. The court imposed a bad-faith litigation penalty (litigância de má-fé) of 5% of the claim value under art. 793-B of the CLT, holding counsel — not the AI — responsible for verifying citations. No OAB disciplinary action is recorded.", "outcome": "sanctions", "sources": [ { "label": "TRT-2 official news release on the ruling (Portuguese) — full acórdão not pulled; process number per secondary analysis", "url": "https://ww2.trt2.jus.br/noticias/noticias/noticia/justica-aplica-ma-fe-em-razao-de-julgados-produzidos-por-inteligencia-artificial" } ], "summary": "In Brazil, the TRT-2 labour appeals court in São Paulo (September 2024) sanctioned a worker's lawyer for bad-faith litigation after an appeal cited AI-fabricated jurisprudence falsely attributed to Superior Labour Court ministers and a non-existent judge. The court imposed a penalty of 5% of the claim value, holding counsel — not the AI — responsible for verifying citations.", "status": "verified" }, { "id": "dayal-fedcfamc2f-1166", "caseName": "Dayal [2024] FedCFamC2F 1166", "date": "2024-08-27T00:00:00.000Z", "court": "Federal Circuit and Family Court of Australia (Division 2)", "jurisdiction": "australia", "aiTool": "LEAP legal-practice-management AI research tool", "conduct": "At a family-law property enforcement hearing, a Victorian solicitor tendered a list of four authorities that did not exist, generated by the AI tool in his practice-management software and never verified. He apologised, paid the other side's thrown-away costs, and Judge Humphreys referred his conduct to the regulator. (Underlying hearing: Handa & Mallick [2024] FedCFamC2F 957.)", "outcome": "other", "regulatoryOutcome": { "body": "Victorian Legal Services Board and Commissioner (VLSB+C)", "disposition": "On 19 August 2025 the VLSB+C varied the solicitor's practising certificate: he may no longer practise as a principal, cannot handle trust money or operate his own practice, may practise only as an employee solicitor, and must undertake two years of supervised practice with quarterly reporting.", "date": "2025-08-19T00:00:00.000Z" }, "sources": [ { "label": "Dayal [2024] FedCFamC2F 1166 (AustLII)", "url": "https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FedCFamC2F/2024/1166.html" }, { "label": "Handa & Mallick [2024] FedCFamC2F 957 — underlying hearing (AustLII)", "url": "https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FedCFamC2F/2024/957.html" }, { "label": "VLSB+C statement — Mr Dayal matter", "url": "https://lsbc.vic.gov.au/news-updates/news/statement-mr-dayal-matter" } ], "summary": "In Australia's Federal Circuit and Family Court, a Victorian solicitor tendered four non-existent authorities generated by the AI tool in his LEAP practice software. Judge Humphreys referred him to the Victorian Legal Services Board and Commissioner, which in August 2025 barred him from practising as a principal and imposed two years of supervised practice.", "status": "verified", "article": "dayal-fedcfamc2f-1166" }, { "id": "zhang-v-chen", "caseName": "Zhang v Chen, 2024 BCSC 285", "date": "2024-02-20T00:00:00.000Z", "court": "Supreme Court of British Columbia", "jurisdiction": "canada", "aiTool": "ChatGPT", "conduct": "Counsel in a family-law matter cited two non-existent cases generated by ChatGPT in a notice of application; the fabrications were discovered by opposing counsel.", "outcome": "costs-order", "regulatoryOutcome": { "body": "Law Society of British Columbia", "disposition": "LSBC investigation into the lawyer's conduct followed the judgment; the court ordered costs against the lawyer personally and a review of her other court files." }, "sources": [ { "label": "Zhang v Chen, 2024 BCSC 285 (CanLII)", "url": "https://www.canlii.org/en/bc/bcsc/doc/2024/2024bcsc285/2024bcsc285.html" } ], "summary": "The first reported Canadian case of AI-hallucinated citations: a Vancouver family lawyer cited two ChatGPT-invented cases in a notice of application, was ordered to pay costs personally, had her other files reviewed, and faced a Law Society of British Columbia investigation.", "status": "verified" }, { "id": "kruse-v-karlen", "caseName": "Kruse v. Karlen, No. ED111172 (Mo. Ct. App., E.D.)", "date": "2024-02-13T00:00:00.000Z", "court": "Missouri Court of Appeals, Eastern District", "jurisdiction": "us-state", "conduct": "A self-represented appellant filed a brief in which 22 of 24 case citations were fictitious AI-generated hallucinations, produced by an out-of-state legal consultant he had hired who used an AI application. The court found the brief flagrantly deficient.", "outcome": "sanctions", "monetaryPenalty": 10000, "penaltyCurrency": "USD", "sources": [ { "label": "Kruse v. Karlen, No. ED111172 (Mo. Ct. App., E.D. 13 Feb 2024) — CourtListener (order PDF)", "url": "https://storage.courtlistener.com/pdf/2024/02/13/molly_kruse_v._jonathan_r._karlen.pdf" } ], "summary": "In the Missouri Court of Appeals, Eastern District, a self-represented appellant's brief contained 22 fictitious citations out of 24, generated by an AI application used by a consultant he had hired. The court dismissed the appeal for briefing deficiencies and ordered him to pay $10,000 in damages for a frivolous appeal.", "status": "verified" }, { "id": "park-v-kim", "caseName": "Park v. Kim, 91 F.4th 610 (2d Cir. 2024)", "date": "2024-01-30T00:00:00.000Z", "court": "US Court of Appeals for the Second Circuit", "jurisdiction": "us-federal", "aiTool": "ChatGPT (OpenAI)", "conduct": "Attorney Jae S. Lee cited a non-existent state-court decision in her reply brief. She admitted using ChatGPT to find supporting precedent and did not read or confirm the fabricated decision before citing it. The court found this fell below the reasonable-inquiry duty under Rule 11.", "outcome": "referral", "regulatoryOutcome": { "body": "US Court of Appeals for the Second Circuit — Grievance Panel", "disposition": "While affirming the dismissal on the merits, the court referred Attorney Jae S. Lee to its Grievance Panel under Local Rule 46.2 for investigation and possible referral to the Committee on Admissions and Grievances.", "date": "2024-01-30T00:00:00.000Z" }, "sources": [ { "label": "Park v. Kim, 91 F.4th 610 (2d Cir. 30 Jan 2024) — CourtListener opinion", "url": "https://www.courtlistener.com/opinion/9469987/park-v-kim/" } ], "summary": "In the US Court of Appeals for the Second Circuit, attorney Jae S. Lee cited a non-existent decision in her reply brief that she obtained from ChatGPT without reading or verifying it. While affirming the dismissal, the court referred Lee to its Grievance Panel under Local Rule 46.2 for failing the reasonable-inquiry duty.", "status": "verified" }, { "id": "harber-v-hmrc", "caseName": "Harber v Commissioners for HMRC, [2023] UKFTT 1007 (TC)", "date": "2023-12-04T00:00:00.000Z", "court": "First-tier Tribunal (Tax Chamber)", "jurisdiction": "uk", "aiTool": "ChatGPT", "conduct": "A litigant in person submitted the names, dates and summaries of nine First-tier Tribunal decisions said to show a reasonable excuse for a late capital-gains disclosure; none of the authorities existed. The tribunal found they had been generated by AI.", "outcome": "dismissal", "sources": [ { "label": "Harber v HMRC [2023] UKFTT 1007 (TC), The National Archives Find Case Law", "url": "https://caselaw.nationalarchives.gov.uk/ukftt/tc/2023/1007" } ], "summary": "The first reported United Kingdom case of AI-fabricated authorities: before the First-tier Tax Tribunal, litigant in person Felicity Harber cited nine non-existent decisions generated by ChatGPT to argue a reasonable excuse for a capital-gains penalty. The tribunal found the cases fake, dismissed her appeal, and warned of the harm fabricated authorities cause.", "status": "verified", "article": "harber-v-hmrc" }, { "id": "parker-v-forsyth", "caseName": "Parker v Forsyth NNO and Others (1585/20) [2023] ZAGPRD 1", "date": "2023-06-29T00:00:00.000Z", "court": "Regional Court, Johannesburg (Gauteng Regional Division)", "jurisdiction": "south-africa", "conduct": "One of the world's earliest reported AI-fabricated-citation cases: plaintiff's counsel is reported to have cited fictitious authorities — fictional names, citations and holdings — apparently produced by an AI chatbot, and the court criticised over-reliance on technology. Details are not yet confirmed against the primary judgment.", "outcome": "other", "sources": [ { "label": "Parker v Forsyth NNO (1585/20) [2023] ZAGPRD 1 (29 Jun 2023) — LawLibrary/Laws.Africa catalogue (neutral-citation authority; judgment body not machine-readable — [NEED: primary text via Juta/LexisNexis])", "url": "https://lawlibrary.org.za/akn/za-gp/judgment/zagprd/2023/1/eng@2023-06-29" } ], "summary": "Widely cited as one of the world's earliest AI-fabricated-citation cases, this Gauteng Regional Court (South Africa) matter reportedly involved counsel citing fictitious authorities produced by an AI chatbot. The neutral citation is corrected to [2023] ZAGPRD 1 (a Regional Court, not High Court, matter); the judgment body remains to be confirmed against a primary source.", "status": "unverified" }, { "id": "mata-v-avianca", "caseName": "Mata v Avianca, Inc.", "date": "2023-06-22T00:00:00.000Z", "court": "US District Court, Southern District of New York", "jurisdiction": "us-federal", "aiTool": "ChatGPT", "conduct": "Attorneys filed an affirmation citing six non-existent judicial decisions generated by ChatGPT, then doubled down when challenged, submitting fabricated opinions.", "outcome": "sanctions", "monetaryPenalty": 5000, "penaltyCurrency": "USD", "sources": [ { "label": "Opinion and Order on Sanctions (Judge P. Kevin Castel, 22 June 2023)", "url": "https://www.courtlistener.com/docket/63107798/mata-v-avianca-inc/" } ], "summary": "The origin case of legal AI hallucination sanctions: two New York attorneys were fined $5,000 after citing six ChatGPT-invented cases in a personal-injury filing and persisting after the fabrications were questioned, prompting the first widely reported judicial rebuke of generative AI misuse by lawyers.", "status": "verified" } ] }