California names expert panel to shape AI oversight and potential frontier model ‘kill switch’

Governor Gavin Newsom’s executive order asks the group to develop recommendations on independent AI auditing, onsite verification and emergency shutdown mechanisms for frontier models

California has appointed four experts to advise on proposals including independent oversight and a potential emergency kill switch for frontier AI models

California Governor Gavin Newsom has appointed four technology and policy experts to advise the state on tighter oversight of frontier AI, including proposals for independent verification inside AI labs and a potential emergency “kill switch” for advanced models.

The advisory group includes Jason Goldman, the first White House Chief Digital Officer; Gillian Hadfield, a professor at Johns Hopkins University whose work includes AI alignment and governance; Alondra Nelson, a professor at the Institute for Advanced Study and former acting director of the White House Office of Science and Technology Policy; and Rob Reich, a Stanford University professor focused on frontier science and technology governance.

Their appointment follows an executive order issued by Newsom on September 18. The order directs California’s Government Operations Agency to accelerate implementation of recently enacted AI oversight laws and, with the Governor’s Office of Emergency Services, develop recommendations for additional safeguards.

California has not enacted a requirement for frontier AI models to include a kill switch. The executive order instead asks the advisory group to examine that mechanism alongside proposals for onsite independent verification.

California considers onsite AI oversight

One proposal would require frontier AI companies to have a designated independent verification organization operating onsite to conduct regular audits and evaluations.

A second would require safety frameworks, transparency reports and risk assessments filed by frontier AI companies under state law to be independently verified against specified standards.

The executive order also calls for work on a potential kill switch that could shut down a frontier model in an emergency, with its effectiveness subject to ongoing independent verification. Another proposal would broaden definitions of critical safety incidents to cover loss-of-control events.

Goldman framed the work around the question of who checks the claims made by AI developers.

“Governor Newsom’s executive order addresses the hardest questions in AI safety governance, including who verifies a frontier lab’s safety claims and whether a model can be reliably shut down,” he says.

Nelson, writing separately on LinkedIn after joining the committee, focused on the institutional mechanisms behind that oversight: “We need durable institutional mechanisms like truly independent verification, auditing and mandatory reporting and the state is leading the way.”

New panel builds on recently signed AI laws

The expert group will be advising against a regulatory framework that California has already begun expanding.

Newsom signed Senate Bill 813 earlier in September, establishing a framework for independent verification organizations to assess AI systems and models for compliance with state law. Assembly Bill 1405 created a state registry for AI auditors and standards covering their independence, transparency and integrity.

Those laws establish the underlying third-party assessment structure. The new executive order directs state agencies to accelerate their implementation while considering whether California should go further with onsite verification and emergency shutdown requirements.

Hadfield says the focus should remain on independent scrutiny of increasingly capable systems: “The need for effective, accountable, and independent oversight of increasingly powerful and autonomous AI is urgent and I’m very glad to join this great team to quickly help ensure we steer in the right direction.”

Reich, who previously served as a senior adviser to the United States AI Safety Institute, said the work will include developing approaches to AI testing and evaluation alongside independent assessment.

The advisory group is expected to develop recommendations for potential changes to California law as the state implements its existing AI audit and verification framework.