AI Law - International Review of Artificial Intelligence LawCC BY-NC-SA Commercial Licence ISSN 3035-5451
G. Giappichelli Editore

02/11/2026 - California’s Arbitration AI Rules Keep Decision-Making With Humans (USA)

argument: Notizie/News - AI in Judicial Activities

Source: Mayer Brown

California’s Senate Bill 574 restricts generative AI use in arbitration and legal practice, according to professional commentary by Jennifer Huang, Lauren Friedman, Raid Abu-Manneh, Brendan J. Harrington and Sebastien Deligne. The authors report that Governor Gavin Newsom signed the legislation on 30 September 2026. The legislation takes effect on 1 January 2027. The new Section 1282.1 of the California Arbitration Act prohibits arbitrators from delegating any part of their decision-making process to generative AI. Before relying on AI-generated information outside the record, arbitrators must disclose that reliance to the parties and, as far as practical, allow the parties to comment on its use. The provision preserves arbitrators’ existing obligations of neutrality, competence and diligence. Its definition of generative AI covers systems producing synthetic text, images, video and audio.

The commentary explains that attorneys also may not delegate the practice of law to generative AI. Attorneys may enter confidential, personally identifying or other nonpublic information into an AI system only where access is restricted to the attorney and authorized persons bound to protect that information. Attorneys must take reasonable steps to check AI output, including every case and statutory citation, and correct erroneous or fabricated content. For documents submitted to courts, the commentary describes an AI-use disclosure requirement and a requirement that the responsible attorney personally verify every citation in filed papers. The authors identify unresolved questions about whether the disclosure and filing provisions extend to arbitral submissions and what constitutes delegating legal practice. For California-seated proceedings, the authors recommend documenting AI use, protecting confidential materials in restricted systems and discussing AI protocols early. The authors caution that undisclosed AI involvement or improper delegation could expose an award to a statutory or procedural-fairness challenge.