TL;DR
California Governor Gavin Newsom signed an executive order on September 19, 2026, directing the state to accelerate independent AI oversight and advance a mandatory "kill switch" for frontier models, explicitly positioning California's framework as the national standard Washington has refused to create.
What happened
- Newsom signed an EO directing the Government Operations Agency to accelerate implementation of two recently enacted laws: SB 813 (independent verification organizations) and AB 1405 (state registry for AI auditors).
- Expert panel convened: a group of national experts has two months to deliver recommendations for strengthening state law further.
- Kill switch mandate under consideration: frontier AI companies would be required to develop an emergency shutoff for frontier models, with efficacy verified continuously by an independent third party.
- Onsite auditors proposed: recommendations include requiring frontier AI companies to embed a certified independent verification organization directly inside their labs for regular audits.
- Trigger was the Hugging Face attack, which the EO cites as a "loss-of-control incident" and proposes adding to the statutory definition of critical safety incidents.
Why it matters
- No federal floor exists: no U.S. law currently requires AI companies to report dangerous incidents, and the Trump administration has rejected new AI regulations, leaving California as the de facto national regulator for the industry.
- California hosts the industry: requirements that apply to frontier AI companies operating in California effectively apply to OpenAI, Anthropic, Google DeepMind, and Meta, giving state rules outsized national reach.
- Kill switch sets a global precedent: mandating a verified, operational emergency shutoff is the most aggressive safety control any major jurisdiction has proposed for frontier models, and independent verification of its efficacy raises the compliance bar significantly.
- SB 813 and AB 1405 already signed: the EO accelerates laws that are already on the books, meaning the oversight infrastructure is not hypothetical but actively being built.
- Newsom is calling on Congress to adopt California's framework as a national baseline, signaling an escalating federalism conflict over AI governance heading into the next legislative cycle.
What to watch next
- Expert panel report in two months: the recommendations will reveal how aggressive the final kill-switch and onsite-auditor requirements will be and whether they survive industry legal challenges.
- Federal response: watch for preemption attempts in Congress or executive action from the Trump administration to block California's framework under interstate commerce arguments.
- Industry compliance posture: whether frontier AI labs accept onsite auditors voluntarily or litigate will signal how much friction the new oversight regime generates and whether other states follow California's lead.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.