72 on the record · +1 this week
Global Legal AI Incident Tracker
Every incident, and what courts and regulators did about it. 72 cases across 13 jurisdictions — each verified against a primary source.
15 Jul 2026 · US-STATE · sanctions
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.
08 Jul 2026 · UK · other
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.
02 Jul 2026 · INDIA · other
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.
01 Jul 2026 · US-STATE · other
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.
03 Jun 2026 · US-FEDERAL · suspension
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.
27 May 2026 · US-STATE · warning
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.
22 May 2026 · UK · referral
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.
guidance diffAUSTRALIA
Australian regulator guidance on generative AI — Queensland's Guidance Statement 37, the three-regulator joint statement, and the NSW solicitor's guide.
analysisUK
England's Divisional Court referred two lawyers to their regulators over AI-fabricated citations and warned its restraint on contempt is not a precedent.
guidance diffCANADA
The version record of Canadian law-society generative-AI guidance — LSBC (2023), the LSO white paper (2024), Alberta's Rules of Engagement, and the CBA toolkit.
analysisUS
Oregon's ~$110,000 AI-sanctions case: terminating sanctions for fifteen fake cases and eight fabricated quotations, and why the headline figure is an aggregate.
analysisAUSTRALIA
A solicitor filed four AI-invented authorities; the court's referral led the regulator to bar him as a principal and impose two years' supervision.