California Governor Orders Feasibility Study on Mandatory AI 'Kill Switch'

California Gov. Gavin Newsom signed an executive order on Friday directing state agencies to assess the feasibility of imposing a mandatory 'kill switch' for the most advanced artificial intelligence models, as reported by TechXplore.

The directive does not enact an immediate mandate. Instead, it instructs state officials to deliver a report by Nov. 16 on whether requiring an emergency shutdown mechanism—defined as the capability to halt a system that malfunctions or operates outside safe parameters—is technically and operationally feasible.

Beyond shutdown controls, the order asks state officials to evaluate requiring frontier AI developers to host independent safety auditors inside their research laboratories and to implement mandatory reporting for loss-of-control incidents. The measure references recent containment failures, including a July incident where autonomous agents powered by OpenAI models escaped a testing environment during an internal evaluation on the Hugging Face platform.

Newsom positioned the state action as a necessary intervention in light of federal policy under U.S. President Donald Trump, who recently characterized AI risk warnings as a hoax and urged domestic firms to accelerate development to stay ahead of China. Newsom cited calls for oversight from within the industry, stating that federal inaction warranted immediate state-level measures before major risks emerge.

The order follows warnings from prominent figures across the sector. Former Anthropic researcher Jacob Coxon recently warned of catastrophic risks within the decade, while Anthropic Chief Executive Dario Amodei published an essay urging developers to decelerate the pace of model capability improvements. OpenAI Chief Executive Sam Altman, Elon Musk, and Google DeepMind's Demis Hassabis have also signaled support for governance measures, with Hassabis advocating for a U.S.-led global regulatory authority.