One page, every item public and dated.
The sanctions
- 23 July 2025. Johnson v. Dunn (N.D. Ala.): three lawyers publicly reprimanded, referred to the bar, and removed from the case over AI-fabricated citations. Their firm had an AI policy; the court found the firm acted reasonably. A policy did not protect the lawyers. (order, Courthouse News; summary, EDRM)
- August 2025 to February 2026. An Am Law 100 firm is involved in three AI-citation incidents in roughly six months, the third after adopting an internal cite-checking policy (the February matter is an accusation, not a ruling). (Above the Law)
- 8 June 2026. A federal judge in Mississippi sanctions four lawyers, two on each side of the same case, over AI-fabricated citations (Withers v. City of Aberdeen, N.D. Miss.). (ABA Journal)
- Running totals. By May 2026, Damien Charlotin's public AI Hallucination Cases database tracked more than 1,000 US court decisions involving AI-fabricated material (GC AI), with new decisions added at several a day (HAQQ); US courts imposed at least $145,000 in sanctions for AI-fabricated citations in Q1 2026 alone (EDRM).
The embarrassments
- October 2025. Deloitte partially refunds the A$440,000 it was paid for an Australian government report after AI-fabricated citations are found (AP, via CTV News); a second incident follows in Canada, with fabricated citations in a health workforce report for Newfoundland and Labrador (The Independent).
The insurance market moves
- 2024. Munich Re's "Mind the Gap" whitepaper predicts AI risk will be excluded because it is difficult to price with traditional means, and asks how a courtroom can determine what an algorithm did when even its creator is unsure. (Munich Re)
- January 2026. ISO generative AI exclusions take effect (CG 40 47, CG 40 48, CG 35 08) (Claims Journal); WR Berkley, Chubb, Travelers, and Berkshire Hathaway constrain AI liability in parallel (CSIS). Meanwhile, in a Geneva Association survey of 600 corporate insurance decision-makers (October 2025), more than 90% said they need insurance cover tailored to AI and generative AI risks (Geneva Association).
- May 2026. EPIC lawyers professional liability survey: 7 of 13 carriers report an increase in AI-related claims. (Insurance Business)
- 10 July 2026. Verisk/ISO reported to be weighing new exclusions for agentic AI. (The Insurer)
- A small affirmative market exists: in April 2025 Armilla launched AI liability insurance underwritten by Chaucer and other Lloyd's underwriters. (Armilla release)
The law arrives
- 16 and 29 June 2026. The EU Digital Omnibus is adopted: high-risk AI duties move to December 2027 and August 2028. (iubenda, Parliament vote; Jones Walker, Council adoption and dates)
- 2 August 2026 (Sunday). The EU AI Act reaches its general application date, and its Article 50 transparency duties take effect: people must be told when they are dealing with an AI, and AI-generated content must be marked. Not postponed. (Jones Walker)
- 2 December 2026. The machine-readable marking grace period for pre-existing generative systems expires. (Jones Walker)