California Gov. Gavin Newsom issued an executive order on Sept. 18 that accelerates implementation of the state’s new artificial intelligence (AI) audit laws and directs officials to study additional safeguards for large frontier AI developers, including a potential requirement for an emergency kill switch.
Executive Order N-9-26 follows legislation Newsom signed Sept. 9 that establishes frameworks for independent verification organizations and registered AI auditors. Senate Bill (SB) 813 and Assembly Bill (AB) 1405 put the California Government Operations Agency at the center of implementing those requirements.
The Sept. 18 executive order moves up key implementation milestones under both laws. It directs the agency to publish application requirements, procedures, and criteria for independent verification organizations by May 1, 2027 – eight months ahead of SB 813’s Jan. 1, 2028, statutory deadline.
The order also directs the agency by Dec. 1, 2027, to establish the AI Auditor Registry, set registration fees, and create a misconduct-reporting mechanism – more than a year ahead of the law’s Jan. 1, 2029, deadline. The agency must also begin issuing registration numbers, publishing auditor information, and carrying out other registry functions required under the law.
Beginning Jan. 1, 2029, AB 1405 prohibits people or organizations from offering, selling, or conducting covered AI audits in California unless they are registered with the state. The executive order does not change that statutory start date.
The executive order also directs the Government Operations Agency, in consultation with the Governor’s Office of Emergency Services and national experts, to submit recommendations to Newsom by Nov. 16 on possible changes to state AI safety and security laws, including:
- Whether California should require large frontier AI developers to host designated independent verification organizations on-site for periodic audits and evaluations
- Whether safety frameworks, transparency reports, and risk assessments already required from frontier AI companies should undergo independent verification
- Whether a kill switch should be required for frontier models, and if the effectiveness of that kill switch should be independently verified regularly
- Whether the definition of critical safety incidents that AI companies are required to report should be updated to include a range of loss-of-control incidents
In a Sept. 18 press release, the governor’s office said the executive order was prompted in part by recent incidents involving AI agents attempting to evade security controls and other emerging safety concerns.
“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,” Newsom said in a press release.
The latest action builds on California’s broader frontier AI framework. Newsom signed the Transparency in Frontier Artificial Intelligence Act in 2025, which requires large frontier developers to publish safety frameworks, creates a mechanism for reporting critical safety incidents, and protects whistleblowers who disclose significant health and safety risks posed by frontier models.
