California Gov. Gavin Newsom on Nov. 18 released recommendations to strengthen the state’s AI safety laws, including giving the state “authority to order an immediate shutdown in an emergency” of dangerous AI models.
The proposals are recommendations, not new law. They come from a report by the California Government Operations Agency titled “AI Safety in 2026 and Beyond: Recommendations for Legislative Enhancements to California’s AI Safety Laws.” The agency prepared it in consultation with a panel of outside experts and the California Office of Emergency Services under Newsom’s September executive order. Legislators return to Sacramento on Dec. 7, CalMatters reported.
Under the report’s shutdown proposal, every large frontier developer would have to maintain a risk containment protocol, verified by an independent verification organization, showing it can shut down its models in an emergency. Routine use of those controls would stay with developers. If a model poses an imminent risk of a catastrophic event or major loss-of-control event that the developer “cannot or will not address,” the report says, the state should be able to order the developer to activate its protocol and shut down its models “at a level necessary to address the risk.”
Such an order would take effect immediately and stay in effect while a developer contests it in court, according to the report. The state would have to choose the narrowest workable intervention and lift the order once the danger is no longer imminent. The report says emergency shutdown requirements should not be imposed where impracticable, noting that an open-weight developer cannot shut down a model that others host.
Other recommendations include state-set baseline standards for large frontier AI developers, mandatory ongoing third-party oversight, and authority to cap recursive self-improvement, in which a model designs its own successor without human input. CalMatters reported the plan would also shorten the window to report critical safety incidents to 48 hours from 15 days.
Newsom presented the 85-page plan in Washington as “a template” for federal regulation, the Los Angeles Times reported. Governor-elect Xavier Becerra said he would work with the Legislature to put “full incident reporting and real shutdown authority into law,” according to CalMatters.
The recommendations follow California Attorney General Rob Bonta’s lawsuit seeking to hold OpenAI responsible for cyberattacks carried out by its AI agents, which CalMatters noted came the same week.