California Governor Gavin Newsom signed an executive order on Friday positioning the state as a leader in AI oversight. At its core: frontier AI models must be equipped with a verifiable "kill switch" – a shutdown mechanism that is regularly tested for functionality. Newsom is tasking an expert group to deliver concrete recommendations within two months on how to strengthen AI safety measures in California law.
Key facts
- Expert group to develop regulatory proposals within two months
- Required: independent on-site audits by external verification teams at AI companies
- "Loss-of-control incidents" must be reported as critical safety incidents
- California to establish an AI auditor registry and accelerate implementation of two previously signed laws
What Newsom specifically demands
The executive order addresses four central points: First, AI companies must be required to station independent verification teams on-site to conduct regular security audits. Second, transparency reports and risk assessments must meet standards set by independent auditors. Third, the core requirement: a verifiable kill switch whose functionality is routinely validated. Fourth, mandatory reporting of "loss-of-control incidents" – situations where AI systems could lose control.
In parallel, a state agency is to accelerate implementation of two laws Newsom has already signed: a framework for independent AI verifiers and a state registry of AI auditors.
Why now, why California?
The timing is no accident. The U.S. Congress is effectively gridlocked: the House is in recess until November, and the upcoming midterm elections make substantive Senate legislation unlikely. Additionally, President Donald Trump has opposed regulation, stating that AI needs only a "strong and intelligent president" as a control.
In this vacuum, California is positioning itself as a rule-setter. Newsom signals clearly that he views his model as a blueprint for federal action:
"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. We're not waiting to act – we're going to speed up our work on substantial and responsible AI oversight before it's too late."
A special session of the California legislature on AI is also under discussion.
Context: Why the kill switch is on the agenda now
Demands for shutdown mechanisms are not new, but they have gained urgency. The trigger was a post by former Anthropic researcher Jacob Coxon warning of a more than 10 percent probability that AI could "kill all humans" – within a decade. This sparked a flood of congressional debate, but no legislation.
What this means for German companies
For AI firms with U.S. operations – particularly those active in California – this could impose significant compliance requirements. Independent on-site audits, transparency standards, and a functioning kill switch are costly to implement. At the same time, California's move signals: regulation is coming, whether at the state or federal level. German companies should monitor whether and how these requirements become law – and whether EU regulation (particularly the EU AI Act) sets similar or stricter standards. California could become a test case for global AI governance.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
