California is advancing AI oversight ahead of federal legislation, establishing new timelines for companies operating frontier models in the state.

Governor Gavin Newsom signed an executive order on Sept. 18, 2026, directing state agencies to speed up two AI oversight laws signed the prior week and to study further safety measures, including a mandatory “kill switch” for advanced models. The order tasks the Government Operations Agency, working with the Office of Emergency Services, to convene outside experts who must deliver recommendations within two months, according to the Governor’s Office release.

The immediate requirement is the expert report, not new rules. The order itself sets no compliance deadline or penalty for enterprises. Proposals under study include onsite third-party verifiers at frontier labs, independently verified kill switches, and expanding “critical safety incidents” to cover loss-of-control events like the cited Hugging Face attack.

For security leaders, the timeline matters more than the order. California already has SB 813 (verification-org certification) and AB 1405 (an auditor registry) in force. Any company running frontier models with a California nexus faces a faster compliance track than peers elsewhere, regardless of headquarters.

With no federal mandate requiring AI firms to report dangerous incidents, California’s expanded definition could become the de facto reporting bar others benchmark against. Newsom is urging Congress and President Trump to adopt the state framework as a national floor, saying “our policy should be the national baseline.”