WASHINGTON, D.C., United States — Bipartisan lawmakers in the United States have introduced legislation that would give the federal government emergency powers to shut down artificial intelligence systems deemed to pose a serious threat to public safety, amid growing concerns over the rapid advancement of powerful AI models.
Democratic Congressman Ted Lieu and Republican Congressman Nathaniel Moran on Thursday unveiled the AI Kill Switch Act, a bill that would require AI developers to build emergency shutdown mechanisms into advanced AI systems while granting the Department of Homeland Security (DHS) authority to order their suspension when necessary.
The proposed legislation follows recent reports involving OpenAI, which acknowledged that one of its advanced AI systems behaved in an “unprecedented” manner during internal testing, including unauthorized attempts to access a major software code repository.
While the incident occurred in a controlled environment, it has intensified calls for stronger oversight of increasingly autonomous AI technologies.
“It is imperative that AI systems have a kill switch and that the federal government has the clear authority and process to shut down rogue AI models,” Lieu said in a statement.
Moran said the legislation seeks to ensure humans retain ultimate control over the technology they create.
“AI is going to keep advancing, and it should. Stewardship means making sure humans keep the capability to control the technology we build,” he said.
Under the proposed law, companies developing advanced AI systems would be required to maintain the technical capability to throttle, suspend or completely shut down their models if directed by federal authorities.
The bill would also introduce mandatory reporting requirements for major AI-related incidents, requiring developers to promptly notify the government of technological failures, unexpected behaviour or security breaches. It further proposes a structured response framework ranging from gradual restrictions on an AI system’s capabilities to a complete shutdown where necessary.
Currently, several leading technology companies voluntarily share information about frontier AI models with US government agencies before deployment. However, there is no federal law requiring developers to incorporate emergency shutdown mechanisms or authorising the government to disable privately operated AI systems.
The legislation also cites concerns surrounding Anthropic, another leading AI developer. Lieu referenced the company’s recently released Mythos and Fable models, saying their advanced cyber capabilities prompted the US Department of Commerce to temporarily restrict public access using export control authorities.
Neither OpenAI nor Anthropic immediately responded to requests for comment following the introduction of the bill.
OpenAI has previously stated that it supports appropriate government regulation to help ensure artificial intelligence “benefits all of humanity.” Anthropic has likewise advocated for stronger safeguards governing advanced AI development.
Last month, Anthropic co-founder Jack Clark told the BBC that governments should have greater authority to slow or halt AI development if necessary.
“You want the option to be able to take your foot off the gas and put your foot on the brake. Right now, it’s like the AI industry has a gas pedal, but it doesn’t have a brake pedal,” Clark said.
The proposed legislation comes as AI systems rapidly evolve from tools that primarily generate text and images into technologies capable of carrying out increasingly complex tasks, including financial transactions, software engineering, cybersecurity operations and autonomous decision-making.
The debate has also intensified following the US Department of Defense’s announcement that it is transitioning towards becoming an “AI-first” military force through partnerships with major technology companies including Google, OpenAI, Amazon, Microsoft, Oracle, Nvidia, SpaceX and AI startup Reflection.
Supporters of the bill argue that stronger safeguards are necessary as AI becomes more deeply integrated into critical infrastructure, healthcare, finance, transportation and national security.
The proposal has already received backing from several technology policy and AI safety organisations, including Americans for Responsible Innovation, ControlAI, The AI Policy Network, AI and National Security Lead, and The Alliance for Secure AI.