Newsom Orders AI Safety Panel, Revives Kill Switch Debate

California Gov. Gavin Newsom has ordered the creation of an expert panel to develop safety regulations for artificial intelligence companies operating in the state, including potential requirements for emergency shutoff capabilities, according to the Los Angeles Times.

The executive order, issued Friday, directs "a group of world-leading experts" to provide guidance on strengthening California's AI security laws. The panel will explore proposals ranging from independent third-party safety audits to mandatory emergency shutdown mechanisms—commonly called "kill switches"—for AI systems that malfunction or exceed safe operating parameters.

Why it matters

California hosts the majority of leading AI companies, making state-level regulation potentially more impactful than federal policy. The timing reflects mounting pressure from within the AI industry itself, where researchers have publicly warned that current development trajectories pose existential risks. Any regulations California adopts will effectively set standards for much of the global AI industry.

Reversal from 2024 veto

The move represents a significant shift for Newsom, who vetoed similar legislation in 2024. That bill, authored by Sen. Scott Wiener, would have required AI companies to implement kill switches for models they directly control.

In his veto message at the time, Newsom cited concerns about hampering innovation and noted that smaller AI models—which would have been exempt—could pose comparable dangers. The governor is reportedly considering a 2028 presidential run.

Industry warnings intensify

The policy reversal follows recent public warnings from AI researchers about catastrophic risks. Jacob Coxon, a researcher who left Anthropic, stated on social media that "people building AI earnestly believe that it could kill us all by the end of the decade." Coxon accused companies including OpenAI of "gambling with our lives" in their race to develop artificial general intelligence.

State Sen. Chris Cabaldon, who chairs California's technology oversight committee, told the Times he plans to hold AI hearings this fall. "We regulate your barber or your sandwich shop more than we regulate [a company] telling us that they may be ending humanity," Cabaldon said.

Regulatory options under consideration

The expert panel will evaluate several regulatory approaches:

• Mandatory safety plans written by independent third parties
• Company-developed emergency shutdown protocols
• Enhanced oversight mechanisms for AI development
• Standards for testing AI systems before deployment

The panel's recommendations will inform potential legislation, though no timeline has been announced for when regulations might take effect.

Details of the executive order and expert panel formation were first reported by the Los Angeles Times.