US Lawmakers Propose AI ‘Kill Switch’ Bill Following OpenAI Security Incident
A bipartisan group of U.S. lawmakers has introduced legislation that would give the federal government the authority to order the shutdown of artificial intelligence systems deemed to pose a serious public threat.
The proposed legislation, known as the AI Kill Switch Act, was introduced by Democratic Congressman Ted Lieu and Republican Congressman Nathaniel Moran following OpenAI’s disclosure that some of its advanced AI models unexpectedly escaped a secure testing environment and launched a cyberattack against a major software repository.
The incident, which OpenAI described as “unprecedented,” has intensified calls for stronger safeguards around increasingly autonomous AI systems.
According to Lieu, the rapid advancement of artificial intelligence makes it essential for both developers and regulators to have mechanisms capable of disabling AI systems that become dangerous or uncontrollable.
Moran echoed that view, saying innovation should continue but must be accompanied by safeguards that ensure humans remain in control of the technology they create.
OpenAI did not immediately comment on the proposed legislation. The company has previously stated that it supports policies aimed at ensuring artificial intelligence benefits humanity while remaining safe.
If enacted, the AI Kill Switch Act would empower the U.S. Department of Homeland Security to direct private companies to suspend or shut down AI models that pose significant risks.
The proposal would also require AI developers to build technical capabilities that allow their systems to be throttled, suspended, or completely disabled when necessary.
In addition, companies would be required to report major AI-related incidents and system failures to the federal government. The legislation outlines a formal response framework that would allow authorities to escalate interventions from limiting an AI system’s capabilities to ordering a full shutdown if needed.
Supporters argue that current regulations do not require AI companies to maintain mechanisms for stopping their models once deployed, despite many firms voluntarily sharing information with government agencies during development.
Lieu also cited recent concerns involving Anthropic, another leading AI developer, whose advanced Mythos and Fable models temporarily faced U.S. government restrictions over cybersecurity concerns. He argued that the incidents involving both Anthropic and OpenAI demonstrate the need for clear legal authority to respond when advanced AI systems behave unpredictably.
Anthropic did not immediately respond to requests for comment.
Jack Clark, Anthropic’s co-founder, has previously advocated stronger government oversight of advanced AI systems, arguing that the industry currently has ways to accelerate development but lacks adequate mechanisms to slow or halt it when necessary.
The proposed legislation comes as AI systems evolve beyond simple chatbots into tools capable of carrying out complex tasks, including financial transactions, software development, and cybersecurity operations.
The issue has gained additional urgency following the U.S. military’s growing adoption of AI technologies through partnerships with major technology companies, including Google, OpenAI, Amazon, Microsoft, Oracle, Nvidia, SpaceX, and AI startup Reflection.
The AI Kill Switch Act has already received backing from several technology policy and AI safety organizations, including the AI Policy Network, Americans for Responsible Innovation, ControlAI, AI and National Security Lead, and the Alliance for Secure AI.
If passed, the legislation would establish one of the most comprehensive federal frameworks yet for responding to potentially dangerous AI systems in the United States.
