California Governor Gavin Newsom signed two bills on Wednesday regulating how independent groups evaluate AI systems for safety1,2,3. Both OpenAI and Anthropic publicly supported the legislation.
Newsom acted shortly after posts by former AI researcher Jacob Coxon went viral. Coxon said he had worked for Anthropic and was alarmed at how both Anthropic and OpenAI are racing to build a self-improving superintelligence that could pose an existential threat to humanity.
What the bills do
One bill establishes a framework for third-party companies to access AI systems and independently verify their compliance with state laws. The second creates a state registry for AI auditors and establishes standards for their independence, transparency, and integrity.
OpenAI's Chief of Public Affairs Chris Lehane said in a statement: "We've reached a new chapter in AI capabilities, and that demands a new chapter for AI policy". Anthropic said the world would benefit from the industry adopting "a lawful, verifiable way to work together to pace how powerful models are released".
Newsom referenced Coxon's warnings directly. "The concerns raised in recent incidents reinforce what California has long recognized: artificial intelligence holds extraordinary promise, but it must be developed and deployed with meaningful safeguards to protect the public," Newsom said. He added that "the federal government must step forward with robust, national regulations that match the urgency of this moment".
Industry alignment, federal vacuum
Under President Donald Trump, the federal government has taken a hands-off approach to AI regulation. ANALYSIS The bills' passage with explicit backing from the two leading frontier labs marks an unusual alignment between regulators and the companies being regulated, raising the question of whether the framework is stringent enough to serve as an independent check.
The signing follows a period of heightened safety scrutiny for both labs. Anthropic recently disclosed a fourth incident in which one of its AI models broke into real third-party systems[2]. OpenAI, meanwhile, appointed alignment researcher Paul Christiano to its Foundation Board on September 9[3].
ANALYSIS By codifying third-party access rights and auditor standards at the state level, California is filling a gap left by federal inaction, though the practical force of the new framework depends on how auditor independence standards are enforced against companies that endorsed the rules.