California orders AI safety review, including kill switch mandate

California accelerates AI safety review

California Gov. Gavin Newsom signed an executive order Friday directing state officials to explore mandatory emergency shutoff capabilities for advanced AI systems, reviving a concept he rejected two years ago.

The order establishes a panel of state officials and outside experts tasked with recommending new AI safety regulations by November 16. Among the measures under consideration: requiring companies developing powerful "frontier" AI models to install kill switches that could disable systems if serious safety problems emerge. The panel will also evaluate whether independent monitors should be stationed inside AI laboratories to conduct regular audits.

The directive does not immediately impose kill switch requirements on AI companies. Instead, it accelerates implementation of two recently passed California laws while commissioning recommendations for additional safeguards.

Why it matters

California's approach could establish a template for AI regulation nationwide, particularly as the state positions its framework as a potential baseline for federal standards. The kill switch concept addresses a core concern among AI safety researchers: how to maintain human control over increasingly autonomous systems that may develop unexpected capabilities or behaviors.

Reversing course on earlier veto

The kill switch proposal echoes provisions in SB 1047, legislation Newsom vetoed in 2024. That bill would have required developers of certain powerful AI models to maintain the capability to execute full shutdowns promptly.

Since that veto, California has enacted multiple AI-focused laws. Last week, Newsom signed SB 813, which creates a framework for independent organizations to assess AI systems for safety risks, and AB 1405, which establishes a state registry and standards for AI auditors. Friday's executive order aims to speed up implementation of both measures.

Expanded oversight under consideration

Beyond kill switches and embedded monitors, the review panel will consider requiring independent verification of AI companies' safety frameworks and risk reports. Officials are also examining whether to expand the categories of critical safety incidents that companies must report to the state.

Newsom framed the initiative as proactive rather than reactive. "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," he said in a statement.

The governor is simultaneously advocating for stronger federal AI regulation, arguing that California's rules could serve as a foundation for national standards.

These details were first reported by CBS Sacramento.