California Governor Gavin Newsom has ordered state agencies to advance proposals for independent oversight of frontier AI and an emergency shutoff mechanism for high-risk models. The executive order does not create an immediate statewide “kill switch,” but directs work on standards, accountability and possible requirements.
The order asks the state to consider whether frontier AI companies should develop an emergency shutoff capability and whether independent organizations should evaluate their safety plans. Any enforceable requirement would need to follow the state’s rulemaking or legislative process.
What California is considering
- Independent third-party assessment of frontier-model safety plans.
- Emergency shutoff capabilities for frontier AI systems.
- Standards for the organizations that verify safety and risk claims.
- Further work on implementation and oversight.
“Kill switch” is shorthand, and its practical meaning matters. A pause mechanism could cover model access, a hosted service or a specific deployment. The governor’s announcement does not identify a single technical design, nor does it say California can remotely switch off every AI model used in the state.
Don’t miss the best of The Mac Observer
Set us as a preferred source and our Apple reporting ranks higher in your Google Search results and Discover feed — one tap, no account changes.
Newsom’s office says the work builds on SB 813, a law establishing a framework for certifying independent verification organizations. The new order aims to accelerate that oversight structure while the state develops a more detailed approach to advanced AI risks.
The order therefore starts a policy process rather than imposing a new consumer-facing rule today. Companies, researchers and the public will have to examine any proposed standards as they are developed.
The Governor’s official announcement sets out the proposals and their current, preliminary status.
Discussion