Labor & Employment Law Daily Wrap Up, INDUSTRY NEWS, TRENDS—California order targets AI oversight, explores frontier-model kill switch, (Sep 21, 2026)

By Patricia K. Ruiz, J.D.

California moved to accelerate AI audits and evaluate emergency shutdown requirements.

California Governor Gavin Newsom issued an executive order directing state agencies to accelerate implementation of California’s new independent artificial intelligence oversight framework and develop recommendations for additional AI safety measures, including a potential “kill switch” for frontier AI models. The order follows recently enacted legislation establishing independent verification organizations to assess AI systems for safety and risk. It also requires state officials to examine whether existing AI laws should be strengthened in response to emerging safety and security concerns.

Need for oversight framework. The executive order, N-9-26, cites reports of attempts to use AI products to create biological weapons, instances in which AI agents allegedly worked to circumvent security controls, and other developments that the administration said raise concerns about threats to public safety. Newsom said California must continue leading both AI innovation and regulation while the federal government has not adopted a comprehensive oversight framework. Newsom urged federal lawmakers to adopt national AI oversight standards, describing California’s regulatory framework as a model for broader regulation of advanced AI systems.

Independent oversight. The order directs the Government Operations Agency to accelerate implementation of recently enacted state laws governing independent AI oversight. By May 1, 2027, the agency must complete application requirements, procedures and criteria for certifying independent verification organizations and publicly post them, as required by law. By December 1, 2027, the agency must also complete additional implementation requirements related to the state’s AI oversight framework and begin carrying out associated statutory duties.

According to the governor’s office, the executive order is intended to speed implementation of Senate Bill 813, which established a framework for certifying independent verification organizations capable of assessing AI systems and models for safety and risk, and Assembly Bill 1405, which created a state registry for AI auditors and established standards governing their independence, transparency, and integrity.

Expert recommendations. The order requires the Government Operations Agency, in consultation with the Governor’s Office of Emergency Services and national experts, to submit recommendations to the governor by November 16, 2026, regarding the technical feasibility and potential effectiveness of amendments to California’s AI safety laws.

Among the proposals identified for review is a requirement that large frontier AI developers host designated independent verification organizations onsite to conduct periodic audits and evaluations. State officials also must assess whether safety frameworks, transparency reports and risk assessments filed by frontier AI companies should be independently verified.

Kill switch proposal. The order specifically calls for recommendations concerning a “kill switch” for frontier AI models and whether such a mechanism could be required under state law. The review is to address the technical feasibility of creating the shutdown capability and whether its effectiveness could be verified on an ongoing basis by an independent verification organization.

The governor’s office said the proposal is part of a broader effort to strengthen safeguards around advanced AI systems. The administration also directed experts to consider expanding the definition of reportable critical safety incidents to include certain loss-of-control events involving AI systems.

Previous action. The order builds on a series of AI-related actions undertaken by California during the past three years. Newsom previously issued executive orders addressing state AI adoption, procurement practices, civil-rights protections, and workforce impacts. The order also references Senate Bill 53, enacted in 2025, which established baseline requirements related to frontier AI safety, transparency, and incident reporting.

News: IndustryNewsTrends Discrimination Safety AINews