Roughly 30 California AI bills go to simultaneous suspense-file votes Thursday, when both the Assembly and Senate Appropriations Committees hold their Aug. 13 hearings. The suspense process is the least legible chokepoint in American AI policy: no recorded debate, no floor speeches, just a chair reading dispositions off a sheet. Bills that die here are dead for the year.

The docket is unusually load-bearing. Senator McNerney’s SB 813 would establish a California AI Standards and Safety Commission and has already cleared the Senate 31-7. Senator Padilla’s SB 903, covering AI in mental-health therapy and transcription, passed 39-0. Padilla’s other bill, SB 867, would ban companion chatbots embedded in toys; it was placed on suspense July 1. Senator Umberg’s SB 574, which Bloomberg Government describes as the first-in-the-nation standard for attorney and arbitrator AI guardrails, cleared the Senate on a bipartisan vote in January.

On the Assembly side: AB 1883 on workplace AI surveillance (approved 52-12), AB 1979 on AI in healthcare services (48-15), and AB 412 on training-data copyright documentation. SB 7 would require 30 days’ notice, with a right to appeal, before any AI-driven hiring, firing, pay or promotion decision.

Zoom out and the stakes get sharper. The Transparency Coalition counts 85 AI-related laws already enacted across 27 states in 2026, with only seven legislatures still sitting. California is the largest of them, and its rulebook tends to become the de facto national one.

Bills that survive Thursday go to floor votes before the Aug. 31 session close, then to Governor Gavin Newsom, who has until Sept. 30 to sign or veto. That’s the same window in which he killed SB 1047 in 2024. The suspense file is where this year’s version of that decision quietly gets pre-made.

Sources

Sources