California SB 574: Senate-passed bill on lawyers’ and arbitrators’ generative AI duties

SB-574 Generative artificial intelligence: attorneys and arbitrators. (2025–2026)

I. Regulatory Summary

If enacted, California legal practices would need a formalized AI-governance layer: approved-tool policies, confidentiality controls, and a repeatable “human verification” process for citations and substantive outputs. Arbitration practitioners should ensure arbitrators’ AI use remains assistive and transparent and does not introduce extra-record information without disclosure.

II. Full Description

The California State Senate passed SB 574 on 2026-01-29, ordering it to the Assembly. The bill would (1) add an express duty for attorneys using generative AI to avoid entering confidential/nonpublic information into public genAI systems; (2) require reasonable steps to verify accuracy, correct hallucinations, and remove biased/offensive/harmful content in AI-generated material used by the attorney; (3) add an explicit rule that court papers must not contain citations the submitting attorney has not personally read and verified, including AI-suggested citations; and (4) restrict arbitrators from delegating decision-making to genAI and from relying on AI-generated information outside the record without disclosure. Status remains proposed pending further legislative action.

III. Scope & Application

Senate passage advances a bill that would codify specific duties for attorneys using generative AI (accuracy verification, confidentiality safeguards, bias/non-discrimination controls) and would add an explicit requirement that court filings not include unverified citations (including AI-suggested citations). It would also restrict arbitrators from delegating decision-making to generative AI and from relying on AI-generated information outside the record without disclosure. Not yet law; further legislative steps remain.

IV. Policy Impact Assessment

Primary Focus: legal ethics and procedural AI safeguards