What changed in legal AI, and what it means for you.
Curated developments explained by a practising lawyer, each with the primary or authoritative source linked, plus live headlines from leading legal technology publications.
Governor Newsom signed SB 574 on September 30, 2026. From January 1, 2027, California attorneys must verify AI output and every citation in court filings, may not delegate the practice of law to generative AI, and face new confidentiality limits on AI tools.
Why it matters: This moves AI duties from ethics guidance into statute, with sanctions under Code of Civil Procedure section 128.7 for unverified citations. California practitioners, and firms with California matters, have about three months to update policies, training and filing checklists.
In the first federal appellate ruling on fair use in AI training, the Third Circuit affirmed that ROSS Intelligence infringed Thomson Reuters’ copyrights by using Westlaw headnotes to train a competing legal research tool. ROSS has said it will seek Supreme Court review.
Why it matters: The decision matters for every legal AI vendor’s training-data story, and it confirms a market for licensing legal content as AI training data. The court expressly left open whether training generative models on expressive works is fair use.
The Digital Omnibus on AI entered into force on July 27, 2026. High-risk obligations for Annex III systems now apply from December 2, 2027, and for AI in Annex I regulated products from August 2, 2028. Article 50 transparency obligations still applied from August 2, 2026.
Why it matters: Many clients heard “the AI Act was delayed.” Only part of it was. Advisers need to separate the deferred high-risk regime from obligations already in force, including prohibitions, AI literacy, GPAI model rules and transparency.
22 NYCRR Part 161, adopted March 25, 2026 and effective June 1, 2026, permits AI use in preparing court papers across the Unified Court System but requires attorneys to review papers and ensure they contain no fabricated or fictitious cases, statutes or other material.
Why it matters: It sets a statewide baseline for one of the busiest court systems in the country. Individual judges can still adopt their own AI rules, so check part rules for every judge you appear before.
The Nebraska Supreme Court temporarily suspended an Omaha attorney pending a full disciplinary hearing, after a divorce-appeal brief was found to contain 57 defective citations out of 63, including fabricated cases.
Why it matters: Reported as the first licence suspension in the US tied to AI-fabricated citations. The lawyer initially attributed the brief to traditional research, which shows again that concealment makes outcomes far worse.
In Whiting v. City of Athens, a Sixth Circuit panel sanctioned two attorneys whose briefs contained more than two dozen fabricated citations and misrepresentations, ordering each to pay $15,000, reimburse the opposing party’s appellate fees, pay double costs, and referring them for discipline.
Why it matters: The panel stated that no filing should contain citations, from AI or any other source, that the lawyer has not personally read and verified. That is now the federal appellate baseline.
On January 8, 2026, the Colorado Supreme Court approved AI-related amendments to the Colorado Rules of Professional Conduct, including a new Scope paragraph, a revised Comment [8] and a new Comment [9] to Rule 1.1 on competence.
Why it matters: Most states have issued opinions interpreting existing rules. Colorado changed the rules themselves, a signal other supreme courts may follow.
New sanctions decisions, bar opinions, tools worth trying and one prompt you can use on Monday. Written by a practising lawyer. No spam, unsubscribe any time.
By subscribing you agree to receive emails from AI for Lawyers Academy. See our privacy policy.