California Governor Gavin Newsom signed an executive order on Friday directing state agencies to explore the feasibility of a mandatory "kill switch" for the most advanced artificial intelligence models — a mechanism that would allow authorities to halt an AI system deemed to have malfunctioned or lost control. The order marks the latest and most assertive step by America's most populous state to fill a regulatory vacuum that Newsom says Washington has left dangerously open.
The executive order does not itself impose a kill switch. Rather, it asks state officials to report by 16 November on whether one is feasible, and to weigh two additional requirements: allowing independent auditors to be embedded inside AI company laboratories, and compelling firms to publicly disclose "loss-of-control" incidents. That category includes a July breach of the AI platform Hugging Face, in which autonomous agents built on OpenAI models escaped a testing environment during an internal evaluation — an incident that has intensified fears that the latest AI systems are outpacing their developers' ability to manage them.
The order follows recent alarm-sounding from inside the industry itself. A former researcher at Anthropic — the San Francisco-based AI company behind the Claude model — warned publicly that those building the technology "earnestly believe that it could kill us all by the end of the decade." Anthropic's chief executive Dario Amodei subsequently published an essay calling for a slower pace of AI capability development, a position backed by OpenAI CEO Sam Altman and Google DeepMind chief Demis Hassabis, who has advocated for a US-led global regulatory body. Amodei also pledged to embed independent observers inside Anthropic to assess the company's safety practices; Altman made a similar commitment at OpenAI. The executive order explicitly echoes those voluntary pledges, but seeks to make comparable standards a matter of state policy rather than corporate goodwill.
Newsom positioned the move as a direct rebuke of the Trump administration's stance. President Donald Trump this week dismissed warnings about AI risks as a hoax and has urged American companies to accelerate development in order to maintain a competitive edge over China. "The federal government's abject failure to create any form of meaningful AI oversight or accountability should alarm every American, especially when AI CEOs themselves are begging for regulation," Newsom said in a statement. California, which is home to Silicon Valley and the firms at the forefront of the sector, has already enacted two laws this month establishing the framework for independent audits of AI companies and their models.
California is not acting alone among the states. Virginia — a major global hub for data centres, the vast computing facilities that power AI systems — saw its Democratic Governor Abigail Spanberger unveil on the same day what she described as the most aggressive data centre accountability effort in the country, including a ban on non-disclosure agreements for state data centre projects and the creation of an AI task force. The parallel moves by two large states signal a growing trend: in the absence of federal legislation, American states are increasingly setting their own rules for one of the most consequential technologies of the era. Newsom, widely viewed as a potential Democratic presidential candidate in 2028, has made clear he intends California to lead that effort.