Summary
On September 18, 2026, California Governor Gavin Newsom signed Executive Order N-9-26, directing the Government Operations Agency to deliver recommendations by November 16, 2026 on four possible amendments to state AI safety law: onsite independent auditors inside frontier labs, independent verification of the safety documents companies already file, a verified "kill switch" for frontier models, and a definition of reportable critical safety incidents that covers loss of control. The order also sets deadlines of May 1, 2027 and December 1, 2027 for the agency to carry out duties under the Government Code, which the governor's office described as accelerating implementation of SB 813 and AB 1405, the independent verification and auditor registry laws Newsom signed nine days earlier. The order itself does not change any legal obligation for AI companies.
What Happened
The order's recitals cite California's earlier AI actions, including Executive Order N-12-23 of September 6, 2023, the frontier AI expert report released in June 2025, SB 53, which took effect in 2026, and Executive Order N-5-26 of March 30, 2026. They then point to "revelations of multiple instances of apparent attempts by individuals to use AI products to create bioweapons and AI agents working, at times independently and at times collectively, to defeat security protocols that AI companies had put in place and working, in some instances undetected for months, to hack other companies," and state that these prompted "some within the AI industry, including company leaders, to call on the industry to pace the development of AI systems and models." The recitals attribute the need for state action to the absence of federal action, "which unfortunately is not forthcoming due to a failure of leadership by the President and Congressional leaders."
The operative text has three numbered orders. The first requires the Government Operations Agency, by May 1, 2027, to complete the requirements of Government Code section 8898.1 and publicly post application requirements, procedures and criteria for independent verification organizations. The second requires it, by December 1, 2027, to complete the requirements of subdivision (a) of section 11549.82 and begin the actions required by subdivision (b). The third directs the agency, in consultation with the Governor's Office of Emergency Services and with national experts, to submit recommendations on "the technical feasibility and potential efficacy" of amending state law to require that large frontier developers "embed designated independent verification organizations onsite in their labs to conduct periodic audits and evaluations"; that the safety frameworks, transparency reports and risk assessments companies file be independently verified; that a "kill switch" be created for frontier models, "with the efficacy of the switch verified on an ongoing basis by an independent verification organization"; and that the definition of critical safety incidents be updated "to include a range of loss-of-control incidents, covering recently reported incidents from large frontier developers."
The governor's press release, headlined as advancing "the creation of an AI kill switch," names the Hugging Face attack as an example of a loss-of-control incident and says the expert group would produce a guide "within two months." It also describes a proposal for "independent third parties to write safety plans for frontier AI companies," which does not appear in the order's list. Newsom said in the release that "the federal government's abject failure to create any form of meaningful AI oversight or accountability should alarm every American, especially when AI CEOs themselves are begging for regulation," and called on Congress and President Trump to adopt California's framework or use it "as a floor, not a ceiling." The release names no experts or institutions for the advisory group.
NBC News described Newsom as "widely seen as a 2028 presidential hopeful," reported that he had raised the possibility of a special legislative session on AI, and noted that Trump has called recent AI safety warnings a "hoax." The same day, Virginia Governor Abigail Spanberger signed an executive order creating an AI task force and a data center accountability plan, and Nevada Governor Joe Lombardo signed an order limiting tax breaks for data centers and requiring subsidized data centers to cover their own power costs.
Why It Matters
The order moves ideas that had been debated at the federal level into a state process with a fixed date: a mandatory shutdown capability for frontier models was proposed in a bipartisan House bill in July 2026, and a Government Operations Agency recommendation is now due to the governor by November 16. It builds on the independent verification organizations created by SB 813, proposing to place them onsite in frontier labs and to have them verify the documents frontier developers already file under state law.
The order is a request for recommendations, not a rule. Any requirement for onsite auditors, a kill switch or a new incident definition would need legislation or rulemaking, and the recommendations have not been published. What a verified "kill switch" would mean technically for a deployed model, and how an outside organization would test one, is not specified in either the order or the release.
§ How to read the metadata
- Landmark
- Fundamentally alters the trajectory; 2–5 per year.
- Major
- Meaningfully shifts the landscape; 2–4 per month.
- Notable
- Worth documenting; significance can be upgraded later.
- Confidence
- High = primary sources corroborate. Medium = credible secondary only. Low = provisional. Disputed = credible sources disagree.
- Contestation
- Uncontested = no formal challenge. Contested = at least one challenge open. Superseded = replaced by a later entry. Unresolved = dispute still open.