California Governor Gavin Newsom has ordered the creation of an expert panel to recommend stronger safety rules for advanced AI systems. The panel will examine measures including a “kill switch” and independent oversight, reflecting concerns that increasingly capable models may evade their intended safeguards or behave unpredictably.

The panel has two months to submit recommendations, which could inform updates to California’s AI laws. Proposed measures include independent organisations carrying out regular audits, companies filing risk assessments with third parties, and reporting “loss-of-control incidents”. The initiative follows similar calls from Anthropic chief executive Dario Amodei for continuing third-party access to evaluate AI safety and compliance.