California's Final-Passage Weekend: 24 AI Bills Race the Aug. 31 Adjournment Clock

California’s Capitol enters its most consequential 72 hours of the AI policy year this weekend. Lawmakers are expected to work straight through Saturday and Sunday ahead of the legislature’s scheduled adjournment sine die at the end of business on Monday, Aug. 31 — and the docket is dominated by artificial intelligence. According to the Transparency Coalition’s Aug. 28 legislative update, three AI-related bills received final approval this week and are on their way to Gov. Gavin Newsom, while 21 more still await full approval by both chambers before the deadline. Any bill that doesn’t clear both houses by noon on Aug. 31 is dead for the session.

The stakes are unusually high this year. Newsom, barred from seeking another term, is widely expected to use his final signing window to burnish a legacy as the most aggressive AI regulator in the United States. Once the legislature adjourns, the termed-out governor will have 30 days to sign or veto everything on his desk — and as the Transparency Coalition’s Bruce Barcott notes, “there’s always a bit of tension, drama, and surprise in September.”

Three bills already on Newsom’s desk

AB 2025 (Pellerin) — Real estate AI disclosure. Requires disclosure whenever AI is used to digitally alter promotional materials in the sale of real property. The bill cleared the Senate 39-1 on Aug. 24 and won Assembly concurrence 78-0 on Aug. 25 — about as close to unanimous as California politics gets.

AB 2392 (Fong) — EdTech procurement standards. Requires the California Community Colleges and California State University systems — and requests the University of California — to convene a joint working group before providing any generative AI system to students, faculty, or staff. The group must deliver procurement standards and training recommendations by Jan. 1, 2028. Passed the Senate 38-0, with Assembly concurrence at 79-0.

AB 2656 (Petrie-Norris) — Union notice for workplace AI. Requires state and local public employers to give written notice to recognized employee organizations at least 45 days before developing, purchasing, or implementing generative AI to perform work within a represented job classification. Senate vote: 39-0; concurrence: 74-2.

They join several AI bills sent to the governor earlier in August, including SB 1159 (Cabaldon), which specifies that under the California Public Records Act and open meeting laws, terms like “person,” “interested party,” and “member of the public” do not include AI systems, autonomous agents, robots, or other nonhuman entities — a quietly significant clarification as agentic AI begins filing records requests and participating in civic processes. SB 928 (Cervantes), also awaiting signature, mandates that California State University instructors must be human, not AI.

The 21 still in play

The pending list reads like a survey of every AI anxiety currently in circulation:

• SB 1119 (Padilla/Wicks/Bauer-Kahan) — The flagship chatbot safety bill, which absorbed its Assembly companion AB 2023. Targets companion chatbots and children’s safety; ordered to third reading Aug. 17.
• SB 503 (Weber Pierson) — Requires developers and deployers of medical clinical decision support systems to identify and mitigate biased impacts. Passed the Assembly Aug. 24 with Senate concurrence Aug. 25 — this one appears headed to the governor’s desk.
• SB 947 (McNerney/Reyes) — Worker protections for AI and automated decision systems, including notice and appeal rights.
• SB 951 (Reyes) — Requires 90-day advance notice before “technological displacement” affecting 25% or more of a covered workforce.
• AB 1979 (Bonta) — AI in healthcare services; amended and passed the full Senate Aug. 24, now awaiting Assembly concurrence.
• SB 1000 (Becker) — Modifies the California AI Transparency Act’s provenance disclosure requirements; passed the Assembly with an urgency clause Aug. 25, now needing Senate concurrence.
• SB 1111 (Ashby) — Clarifies that false impersonation law covers digital replicas used to defraud.
• SB 300 and SB 867 (Padilla) — Companion chatbot restrictions: blocking sexually explicit content from companion bots, and prohibiting companion chatbots in toys outright.
• SB 813 (McNerney) — Establishes a California AI Standards and Safety Commission, approved 11-3 out of Appropriations and ordered to third reading Aug. 26.
• SB 903 (Padilla/Rubio) — Restricts AI transcription of patient information in mental health therapy.

Why this weekend matters

California remains the de facto national laboratory for AI regulation. With Congress still unable to pass comprehensive federal AI legislation — and the June executive order pushing a light-touch, innovation-first posture — states have filled the vacuum. The Transparency Coalition counts 85 new AI-related laws enacted across 27 states in the first half of 2026 alone. But California’s scale, its concentration of AI companies, and its precedent-setting habit mean that what survives this weekend’s floor votes will shape compliance roadmaps far beyond Sacramento.

Two patterns stand out in the final sprint. First, bipartisanship: nearly every bill moving this week passed with lopsided or unanimous margins, suggesting AI disclosure and safety provisions have escaped the partisan framing that stalled earlier, more sweeping proposals like SB 1047 in 2024. Second, a shift from frontier-model regulation toward application-specific rules — real estate ads, university procurement, union notice periods, therapy transcription, toy chatbots. After the fights over developer liability killed earlier omnibus bills, legislators have learned to regulate AI where it touches specific constituents rather than at the model layer.

Watch the floor calendars through Monday night. The chatbot safety package (SB 1119), the healthcare AI cluster (AB 1979, SB 503), and the labor notice bills (SB 947, SB 951) are the most consequential measures still hanging in the balance. Whatever clears both chambers lands on a governor with 30 days, a legacy to define, and national ambitions that may hinge on being the politician who took AI seriously first.