California: independent verifiers and a register of AI auditors
On 9 September 2026 California's governor Gavin Newsom signed two acts. SB 813 requires the Government Operations Agency to set, by 1 January 2028, the procedure by which an AI auditor with demonstrated expertise in assessing risk is designated an independent verification organization. AB 1405 creates, by 1 January 2029, a state register of AI auditors, after which an unregistered person may not conduct such audits.
Why it matters
An outside verifier of AI systems became in law a status granted by the state, with independence requirements, registration and a duty to report. Until then outside evaluators worked under agreements with the labs themselves, and the lab decided what of their findings to publish.
SB 813 (by Senator Jerry McNerney) is Chapter 179 of the state statutes, AB 1405 (by Assemblymember Rebecca Bauer-Kahan) Chapter 178. A verification organization reports to the agency and the Legislature each year, but no sooner than 12 months after designation. A registered auditor may not negotiate employment with the auditee while the audit is under way. On 18 September executive order N-9-26 told the agency to speed up both acts and convened experts to propose within two months how to strengthen state law, including requiring an independent verifier to be embedded in frontier labs and a kill switch for frontier models; the order cites recent incidents, among them the Hugging Face attack. On 23 September the governor named four experts. What the record does not claim: that any organization has yet been designated or registered, or that the acts oblige labs to hire a verifier. The session did not open the signed text of the order; its content is given from the governor's release.