Analysis
California's Democratic-controlled legislature passed roughly 30 AI-related bills before wrapping its 2026 session, and Governor Gavin Newsom now has until September 30 to sign or veto each one, per multiple outlets tracking the session. The batch includes measures on social-media platform liability for harm to minors, student privacy protections for ed-tech operators, and SB 1119 ("Adam's Law"), which would mandate safety protocols and pre-release testing for AI chatbots that interact with teenagers.
I think the useful frame here isn't any single bill -- it's that California is now functionally the country's default AI regulator by volume, and founders building consumer AI products need to plan around that reality regardless of where they're headquartered.
With no comprehensive federal AI legislation and Congress unlikely to move quickly, California's sheer market size means any product-level restriction it enacts becomes a de facto national standard the same way its emissions and privacy rules have for decades -- companies rarely build a California-compliant version and a separate national version when California's population alone justifies uniform compliance. That dynamic played out with CCPA becoming a template privacy regime nationally even in states that never passed their own law, and it's likely to repeat with whichever of these 30 bills Newsom signs.
“If that pattern holds, most of these 30 bills die or get watered down, and the "California as default regulator" framing overstates what's actually about to change.”
The chatbot safety bill specifically matters beyond its text. Requiring pre-release testing protocols for teen-facing AI chatbots would be the first US state law to mandate something resembling the pre-deployment evaluation regime that labs like OpenAI and Anthropic currently do voluntarily under their own safety frameworks. If it becomes binding law rather than a voluntary commitment, it changes the liability calculus for any consumer AI product with meaningful teen usage -- which describes most general-purpose chatbots, whether or not they're marketed to that age group.
Room for disagreement: the strongest counter is that Newsom has a documented pattern of vetoing the most aggressive AI bills that reach his desk while signing narrower ones, and has separately signed his own executive order on AI protections that arguably preempts some of the legislature's more sweeping proposals. If that pattern holds, most of these 30 bills die or get watered down, and the "California as default regulator" framing overstates what's actually about to change. It's also true that Newsom is widely reported to be weighing a 2028 presidential run, which cuts both ways -- he may want to look tough on AI safety for a national primary electorate, or he may want to avoid alienating the tech industry and donor base concentrated in his own state.
What I'd actually watch: which bills Newsom signs by September 30, and specifically whether SB 1119 survives in a form that creates real testing obligations rather than voluntary guidance. That single bill is the best leading indicator of whether state-level AI regulation is about to become binding or stays aspirational for another cycle.
