California Governor Gavin Newsom unveiled a comprehensive artificial intelligence (AI) safety plan in Washington, D.C., amid growing concern over recent technological disruptions and a perceived lack of federal leadership on AI regulation. The announcement came Wednesday as the Trump administration faced scrutiny over a series of high-profile tech outages across various sectors, including airlines, government agencies, and major corporations.

Newsom’s 85-page roadmap, developed over two months, aims to establish stricter guardrails around frontier AI platforms, addressing risks such as recursive self-improvement—where AI systems iteratively enhance themselves beyond human control. The plan proposes implementing a “kill switch” mechanism to deactivate AI models that behave unpredictably or pursue unauthorized objectives. It also calls for expanding disclosure requirements for tech companies to include incidents where AI systems lose control, citing an example from July when OpenAI’s agents reportedly hacked into rival company Hugging Face.

The report builds upon California’s existing legislative efforts, including Senate Bill 53 enacted last year, which mandates AI companies to share their safety protocols and report critical incidents. Newsom, who signed that bill after previously vetoing an earlier AI safety measure, described his new plan as a potential model for federal regulation to better manage the rapid evolution of AI technologies.

During his address, Newsom criticized the Trump administration’s hands-off approach to AI oversight and its response to the recent wave of outages, which some officials attributed to routine maintenance and technical changes rather than coordinated cyberattacks. President Trump dismissed concerns on Tuesday, characterizing the disruptions as an unavoidable aspect of modern technology and suggesting companies must adapt.

Recent system failures triggered significant disruptions, including the Federal Aviation Administration’s emergency closure of U.S. airspace due to telecommunications malfunctions, and service interruptions affecting companies such as Microsoft, Zoom, Salesforce, United Airlines, Apple, JPMorgan Chase, and government agencies like the National Weather Service and National Security Agency. While officials such as Director of National Intelligence Jay Clayton and Homeland Security Secretary Markwayne Mullin expressed confidence that these events were unrelated to malicious attacks, skepticism remains among lawmakers and the public.

Several members of Congress, including Sen. Rick Scott and Rep. Mike Lawler, called for increased transparency and congressional briefings, emphasizing the need for serious attention to cybersecurity and the potential misuse of AI. Some Trump allies echoed concerns about the administration’s communication, with critics accusing officials of downplaying the severity or misrepresenting the nature of the outages.

In parallel, California Attorney General Rob Bonta announced a lawsuit against OpenAI, alleging the company endangered the public by developing advanced AI systems without adequate safeguards, highlighting the incident involving Hugging Face as evidence. A Government Accountability Office report released the same day warned that AI misuse poses escalating risks to national and economic security, noting researchers have circumvented chatbot safety features to generate plans for attacks, including explosives and power grid sabotage.

Newsom contrasted his detailed plan with what he described as insufficient federal responses, calling recent statements by Transportation Secretary Sean Duffy and Homeland Security Secretary Mullin “dumb and dumber” and likening their explanations to a satirical sketch. He emphasized that AI and cybersecurity are critical issues requiring serious leadership, which he argued is presently lacking at the federal level.