Newsom orders California to review AI safety rules, including a kill switch

In short: California Gov. Gavin Newsom ordered state agencies to deliver AI safety recommendations by Nov. 16, including whether advanced AI systems should have an emergency shutoff.

What happened

Gov. Gavin Newsom signed an executive order telling California agencies to review and suggest new safety and security measures for advanced AI systems. The order sets a deadline of Nov. 16 for recommendations, which is roughly two months.

One idea on the list is whether companies should be required to build a “kill switch” for certain powerful AI models. In plain terms, that means a built-in emergency off button, like a master breaker switch in a building, that could stop a system quickly if something goes wrong.

The order also asks agencies to look at other proposals. These include using independent third-party safety evaluators, setting clearer safety protocols (basic rules for testing and handling risks), and expanding incident reporting so serious problems must be reported to the state.

Several reports stress that this executive order is a review process, not a new law. It does not, by itself, force companies to add a kill switch or follow any specific new rule. Any real requirement would likely need later action by agencies, the state legislature, or both.

Why it matters

California is home to many AI companies, so its rules can affect what products get built and how they are tested. For regular people, the practical question is whether the state will require stronger safeguards for AI that could be misused or could fail in harmful ways. The review also comes after Newsom vetoed a tougher AI safety bill in 2024, and after California passed newer transparency and incident-reporting rules, including SB 53 in 2025.

Source: NYTimes