California Executive Order N-9-26 on Independent AI Oversight and an AI Kill Switch
Issued by
California Governor's Office
- September 30, 2026 · Correction — Rewrote the entry from the signed order (N-9-26): it asks experts for recommendations within two months and imposes no duties, enforcement, or shutdown powers on companies (fact-check finding). (Cody Maxwell)
- October 1, 2026 · Correction — Typed as an executive order, a new directory category. (Cody Maxwell)
- October 1, 2026 · Correction — Rewrote the practical steps, newsletter hook, search description, and audience fields to match the entry's corrected content. (Cody Maxwell)
Governor Newsom signed Executive Order N-9-26 on 18 September 2026. It convenes experts to recommend, within two months, changes to California's frontier AI safety laws, including independent verification and an emergency shutoff for frontier models. It imposes no new duties on companies itself.
Applies To
Overview
Governor Newsom signed Executive Order N-9-26 on 18 September 2026. It directs the Government Operations Agency to work with the Governor's Office of Emergency Services and outside experts. Within two months they must recommend how to strengthen California's AI safety and security laws. The experts are to consider several proposals for large frontier AI developers. These include onsite independent verifiers, third-party verification of safety disclosures, and an emergency shutoff, or kill switch, for frontier models. They also include widening critical incident reporting to cover loss-of-control incidents. The order also asks the agency to move ahead with its existing duties under SB 813 and AB 1405, the 2026 laws on AI auditors. It creates no new requirements for private companies, no enforcement powers, and no shutdown authority. Any new duties would need legislation.
Key Requirements
- •Directs the Government Operations Agency, with Cal OES and outside experts, to recommend changes to state AI safety law within two months.
- •Proposals to study include onsite independent verifiers at large frontier developers and third-party checks of safety disclosures.
- •Experts will also consider an emergency shutoff for frontier models and broader loss-of-control incident reporting.
- •Asks the agency to move ahead with existing SB 813 and AB 1405 duties.
- •Imposes no new requirements, enforcement powers, or shutdown authority on private companies.
What Your Organization Must Do
- →Track the expert recommendations due around mid-November 2026 for proposals affecting frontier AI developers.
- →Brief executives that this order imposes no new company duties and no shutdown authority.
- →Review existing obligations under SB 813 and AB 1405 on AI auditors, which the order tells agencies to advance.
- →Assess whether your frontier models could support an independently tested emergency shutoff, in case legislation follows.
- →Map current critical incident reporting to see if it would capture loss-of-control events.
- →Prepare engagement with California agencies while these proposals are still under study.
Playbook Guidance
Step-by-step implementation guidance for compliance teams.
Frequently Asked Questions
- Does the California AI kill switch executive order apply to private companies not contracting with the state?
- Not directly: the order tasks state agencies and experts with recommendations, so new duties for companies would need legislation.
- What qualifies as a high-risk AI system under the California AI oversight executive order?
- The order does not define high-risk AI. Its proposals concern large frontier AI developers, as defined in existing California law.
- Does the order require companies to build a kill switch?
- Not yet: the order asks experts whether to require an independently tested emergency shutoff for frontier models.
- When does the California AI oversight executive order take effect and what is the timeline?
- The order was signed and took effect on 18 September 2026. Expert recommendations are due within two months, around mid-November 2026.
- Can California state agencies compel shutdown of a private company's AI system under this executive order?
- No, the order creates no shutdown or enforcement power. It only asks for recommendations and faster work on existing laws.
- How does the California AI kill switch order compare to EU AI Act human oversight requirements?
- The EU AI Act already requires human oversight for high-risk systems. The California order only studies a possible kill switch for frontier models, so nothing comparable is yet required.