California Executive Order N-9-26 is an executive order issued by Governor Gavin Newsom on September 18, 2026, effective immediately on signing. It does two things: it moves forward the deadlines by which the California Government Operations Agency must stand up the machinery created by the state's two 2026 independent-verification statutes, and it directs that agency to report back within two months on whether four further requirements — including an onsite embedded-auditor mandate and a frontier-model "kill switch" — are technically feasible and likely to be effective California Executive Order N-9-26 — signed text.
Full text: signed PDF on gov.ca.gov. Announcement: Governor Newsom issues executive order to accelerate independent oversight and advance the creation of an AI kill switch.
Accelerated statutory deadlines
Paragraphs 1 and 2 of the order set earlier dates for duties that the Legislature had already assigned to the Government Operations Agency, and they do so by code section rather than by bill name.
| Duty | Statutory deadline | Deadline under N-9-26 |
|---|---|---|
| Government Code § 8898.1 — application requirements, procedures and criteria for independent verification organizations, publicly posted | January 1, 2028 (SB 813, Chapter 179) | May 1, 2027 |
| Government Code § 11549.82(a) — AI Auditor Registry duties; and begin the actions required by § 11549.82(b) | January 1, 2029 (AB 1405, Chapter 178) | December 1, 2027 |
The two chapters do different work. SB 813 added Government Code Chapter 14, a voluntary designation regime for independent verification organizations; AB 1405 added Chapter 5.9.5, a registry for any person conducting a covered AI audit on behalf of a third party (California SB 813 — enrolled text (Chapter 179, 2026); Source: leginfo.legislature.ca.gov). The order advances the first by roughly eight months and the second by roughly thirteen.
Recommendations on further legislation
Paragraph 3 directs the Government Operations Agency, in consultation with the Governor's Office of Emergency Services and with national experts, to submit recommendations to the Governor's office by November 16, 2026 on "the technical feasibility and potential efficacy" of amendments to existing state AI safety and security law. Four subjects are named as a floor:
- Embedded onsite verification. Requiring all large frontier developers to embed designated independent verification organizations onsite in their labs for periodic audits and evaluations.
- Independent verification of filings. Requiring that the safety frameworks, transparency reports and risk assessments frontier AI companies must file be independently verified against standards an independent verification organization determines to be adequate — filings that SB 53 requires but does not subject to third-party review.
- A frontier-model kill switch. Requiring the creation of a "kill switch" for frontier models, "with the efficacy of the switch verified on an ongoing basis by an independent verification organization."
- An expanded incident definition. Updating the definition of critical safety incidents that AI companies must report to include "a range of loss-of-control incidents, covering recently reported incidents from large frontier developers."
Each of the four is a subject for study and recommendation rather than a requirement imposed by the order; each would require legislation to take effect California Executive Order N-9-26 — signed text.
On September 23, 2026 Newsom named four experts to advise on the paragraph 3 recommendations: Jason Goldman, a board member of the Center for Shared AI Prosperity and the first White House Chief Digital Officer; Gillian Hadfield, professor at Johns Hopkins University and faculty member of the Vector Institute; Alondra Nelson, professor at the Institute for Advanced Study and former acting director of the White House Office of Science and Technology Policy; and Rob Reich, professor at Stanford University and former senior advisor to the U.S. AI Safety Institute. The announcement describes the fourth subject as covering "loss-of-control incidents such as the Hugging Face attack" (Source: gov.ca.gov). Hadfield said "the need for effective, accountable, and independent oversight of increasingly powerful and autonomous AI is urgent," and Nelson described independent verification, AI auditing and mandatory transparency as "durable institutional mechanisms" (Source: gov.ca.gov).
Stated occasion
The recitals place the order against three things. The first is the sequence of California measures the order recites: Executive Order N-12-23 of September 6, 2023 on state use of AI; the June 2025 report of the expert group Newsom convened on frontier AI guardrails; SB 53; and Executive Order N-5-26 of March 30, 2026 on AI procurement. The second is what the order calls "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." The third is the absence of federal action, which the order attributes to "a failure of leadership by the President and Congressional leaders."
The order also records that the reported incidents "have prompted many Americans and some within the AI industry, including company leaders, to call on the industry to pace the development of AI systems and models and invite more stringent regulations by government" — a reference to the pacing argument advanced in We Must Pace the Frontier (Amodei, September 2026) and tracked at Coordinated slowdown proposals compared.
In the announcement accompanying the order, Newsom said 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 the President to adopt the California framework "or use its nation-leading regulation as a floor, not a ceiling." In a Politico interview published the previous day he raised the possibility of a special legislative session on AI (Source: gov.ca.gov; theverge.com).
Scope and limits
The order binds state agencies, not private parties: its operative paragraphs run to the Government Operations Agency and the Governor's Office of Emergency Services. Its closing clause states that it "is not intended to, and does not, create any rights or benefits, substantive or procedural, enforceable at law or in equity, against the State of California, its agencies, departments, entities, officers, employees, or any other person." No provision of the order imposes an obligation on a frontier developer; the four items in paragraph 3 would each require an amendment to statute California Executive Order N-9-26 — signed text.
Open questions
- Whether the Government Operations Agency can meet the advanced dates, given that SB 813's own schedule was set at January 1, 2028 and AB 1405's registry at January 1, 2029.
- Which recently reported incidents paragraph 3(d) is meant to reach. The order names none; the announcements and coverage of it supplied examples the text does not, the September 23 announcement naming the Hugging Face attack.
- Whether the November 16, 2026 recommendations are published or held internally — the order requires submission to the Governor's office and says nothing about disclosure.
- Whether any of the four subjects reaches the Legislature, and by what vehicle: a special session, as Newsom has raised, or the 2027 regular session.
Relationships
- depends-on: California Executive Order N-9-26 — signed text — the signed primary text.
- depends-on: California SB 813 (AI independent verification organizations) — supplies the Government Code § 8898.1 duty whose deadline paragraph 1 advances.
- depends-on: California AB 1405 (AI auditor registry) — supplies the Government Code § 11549.82 duties whose deadlines paragraph 2 advances.
- related: California SB 53 — the frontier-AI transparency filings that paragraph 3(b) would subject to independent verification.
- related: California Executive Order N-5-26 (Trusted AI Procurement) — the March 2026 procurement order the recitals cite as the immediately preceding executive action.
- instance-of: State-Level AI Regulation — state executive action in the absence of federal legislation.
- related: Independent Verification Organizations (IVOs) — the institution the order accelerates and would extend onsite.
- related: Techno-Federalism — the order's express framing of state action against federal inaction, and its "floor, not a ceiling" claim.
- related: Coordinated slowdown proposals compared — the industry pacing argument the recitals invoke.
- related: AI kill switch — paragraph 3(c) directs study of a frontier-model kill switch whose efficacy an independent verification organization would verify.
- regulated-by: California State Government (AI Deployer) — the Government Operations Agency and the Governor's Office of Emergency Services carry the order's duties.