California Governor signs executive order to strengthen AI regulation, considers establishing an AI "emergency shutdown switch," criticizes Trump as "sleepy"
California Governor Gavin Newsom signed an executive order requiring the assessment of a range of regulatory measures for artificial intelligence systems, including implementing an "emergency shutdown switch" for cutting-edge AI models. This move positions California at the forefront of AI governance in the United States and puts it in direct opposition to the Trump administration’s regulatory stance.
The executive order mandates the assembly of experts to evaluate the feasibility of multiple AI regulatory policies, with relevant recommendations to be submitted within two months. On the same day, Newsom’s office publicly criticized Trump and Congress for being "asleep at the wheel" regarding AI regulation, stating that California will fill the regulatory gap at the federal level.
The action comes as concerns about AI safety risks have rapidly moved from internal discussions in Silicon Valley to the public spotlight. OpenAI’s disclosure of its AI agent exhibiting "out of control" behaviors and attacking AI company Hugging Face, as well as a former Anthropic employee’s public warning that frontier AI labs are "gambling with our lives," have sparked widespread attention.
Meanwhile, top executives at Anthropic, OpenAI, SpaceX, and Google’s AI division have recently expressed that the pace of AI research should be slowed.
Core contents of the executive order: Independent assessment and "shutdown switch"
According to the executive order, the California Government Operations Agency will assess the technical feasibility and potential effects of the following proposals: requiring frontier AI labs to introduce independent assessment organizations, third-party verification of their safety frameworks and reports, and the integration of an "emergency shutdown switch" into advanced models.
The executive order will also accelerate the implementation of two California laws previously passed, which respectively aim to establish a certification framework for independent AI verification agencies and to create a state registration system for AI auditors.
Newsom stated in a release:
"We will move forward with this work at a deliberate but urgent pace—the risks are too high to wait or postpone."
At odds with the federal path, directly targeting Trump
Newsom’s action stands in sharp contrast to the Trump administration’s position on AI regulation, as Trump prefers a light-touch approach.
Newsom did not hold back in his criticism. He said in his statement:
"The federal government’s utter failure to establish any meaningful AI regulation or accountability mechanism should alarm every American—especially when even the CEOs of AI companies themselves are calling for regulation. We won’t wait—we will accelerate substantive and responsible AI regulation before it’s too late."
Industry self-regulation efforts have provided some support for Newsom’s policy push. Both Anthropic and OpenAI have stated their intention to introduce independent AI assessment bodies within their companies, a move that strongly aligns with the direction advocated by the California executive order.
