California sends a wave of AI bills to Newsom as legislative session closes
Real-estate image disclosure, workplace AI notice and medical-AI bias-mitigation bills passed within a day of each other; Newsom has until 30 September to act on each.
- Government & policy
- Notable
As California’s two-year legislative session closed on 31 August 2026, lawmakers sent a cluster of AI-specific bills to Governor Gavin Newsom’s desk within the space of three days. Three completed passage on 25 August: AB 2025 requires landlords and their agents to disclose when rental-listing photos have been substantively altered by photo-editing software or AI, and to provide a link to the unaltered image, exempting routine adjustments such as cropping or colour correction; AB 2656 requires public employers to give recognised employee unions 45 days’ written notice before deploying generative AI to do work covered by a union contract; and SB 503 requires developers and deployers of AI “clinical decision support” tools — software that recommends diagnoses or treatment — to identify and mitigate risks of biased outcomes and document the demographics of their training data.
Two more followed within 48 hours. AB 2392, passed 26 August, directs California’s public university systems to jointly set procurement standards for generative AI tools by January 2028, covering bias assessments, limits on “sycophantic” behaviour and vendor transparency. SB 1000, which completed passage on 27 August carrying an urgency clause, expands California’s existing AI-content-disclosure law, requiring covered providers to offer free tools letting the public check whether content was AI-generated or altered and to embed permanent provenance data — provider, system name and timestamp — inside AI-generated media.
Because all five bills reached the governor in the final days of the two-year session, they fall under the state’s end-of-session rule: rather than the usual 12 days, Newsom has until 30 September to sign or veto each one.
The wave addresses narrower, consumer- and public-sector-facing applications of AI rather than the frontier-model safety obligations that dominated California’s two previous major AI fights — Newsom’s 2024 veto of SB 1047 and his signing of the narrower SB 53 a year later — and sits alongside a May 2026 executive order directing state agencies to prepare for AI-driven workforce disruption, continuing a pattern of regulating specific AI applications and harms rather than model capability itself.