California Governor Newsom proposes mandatory regulation of frontier AI and grants emergency shutdown authority
纽森提议对前沿人工智能实施强制监管并赋予紧急关停权
California Governor Newsom issued legislative proposals on November 18, 2026, that would require continuous independent oversight of large frontier AI developers and authorize state officials to order emergency shutdown of models posing an imminent catastrophic risk. The proposal is based on the report "AI Safety in 2026 and Beyond" released by the state Government Operations Agency on November 16, and is for lawmakers to consider when they reconvene in December; it has not yet become a mandatory obligation. Under the proposal, large frontier developers must undergo continuous oversight by licensed independent verification organizations (IVOs) and establish risk-containment protocols certified by executives and tested by IVOs; if state officials have credible reason to believe there is an imminent catastrophic risk and the developer refuses containment, they may order immediate shutdown. The incident reporting deadline would be shortened from 15 days to 48 hours, with a 24-hour requirement retained for imminent catastrophic risk; models with catastrophic action capabilities are treated as frontier models for incident reporting even if trained with less than 10^26 compute. The proposal responds to Newsom's September 18 executive order on independent oversight and an AI emergency shutdown switch.