California orders work on an AI “kill switch”

Governor Gavin Newsom signed an executive order on September 18 directing the Government Operations Agency, with the Office of Emergency Services, to convene national experts within two months and come back with recommendations for strengthening California's AI laws — accelerating the timelines in SB 813 and AB 1405 along the way.

Four things are explicitly on the table: independent third parties stationed on site inside frontier labs for regular audits, outside verification of the labs' safety frameworks and risk assessments, an emergency shutoff for frontier models whose effectiveness is re-checked over time, and a rewrite of what counts as a critical safety incident to include loss-of-control events.

The order cites this summer's Hugging Face attack as the reason for moving now, and asks Congress and the President to take California's framework as a national baseline — a direct answer to federal efforts to preempt state AI rules. “We're not waiting to act,” Newsom said. “We're going to speed up our work on substantial and responsible AI oversight before it's too late.”

• Confirmed Under consideration: independent auditors stationed inside frontier labs, verification of safety frameworks, and an emergency shutoff whose effectiveness is checked over time. Governor of California
• Confirmed The order also updates critical safety-incident definitions to include loss-of-control incidents. Governor of California
• Confirmed Newsom: “We’re not waiting to act – we’re going to speed up our work on substantial and responsible AI oversight before it’s too late.” Governor of California