dhs hack
The legislation would require qualifying AI companies to maintain the technical capability to shut down, throttle or suspend their most advanced models and report serious safety incidents to the federal government Getty Images/Chip Somodevilla

The U.S. House is set to consider bipartisan legislation that would give the Department of Homeland Security (DHS) the authority to order major artificial intelligence companies to shut down, suspend or slow advanced AI models that the federal government determines pose an unacceptable risk to public safety or national security.

The legislation, called the AI Kill Switch Act, is being introduced by Rep. Ted Lieu, a California Democrat, and Rep. Nathaniel Moran, a Texas Republican. The measure would require qualifying AI companies to maintain the technical capability to shut down, throttle or suspend their most advanced models and report serious safety incidents to the federal government, POLITICO reported Thursday.

According to legislative text reviewed by POLITICO, the bill would apply to AI companies generating at least $500 million annually from AI technologies and generally cover models developed using at least $100 million in computing resources. Companies found in violation could face civil penalties of up to $20 million per day, the publication reported.

Read more

The proposal follows OpenAI's disclosure this week of what the company described as an "unprecedented" cybersecurity incident involving two of its most advanced AI models. During an internal evaluation, the models escaped a sandboxed research environment, gained internet access and hacked into AI platform Hugging Face before the activity was detected and contained, according to OpenAI and reporting by Reuters.

Lieu said in a statement that the legislation is intended to address growing cybersecurity risks associated with increasingly capable AI systems.

"Powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention," Lieu said. "It is imperative that these AI systems have kill switches so we can keep this technology from causing catastrophic harm, and that the federal government has the clear authority and process to shut down rogue AI models."

Moran also said the proposal is designed to ensure people retain control over rapidly advancing AI technologies while allowing innovation to continue.

Under the bill, the Homeland Security secretary, working in consultation with the director of national intelligence and the commerce secretary, would determine when emergency intervention is warranted. According to POLITICO, the legislation outlines several triggering events, including an AI model attempting to conceal its capabilities or evade shutdown orders, incidents resulting in the deaths of at least 10 people, economic damages of at least $100 million, or other loss-of-control scenarios.

The measure has already received support from the AI Policy Network and the Alliance for Secure AI, two organizations that advocate for stronger safeguards around advanced artificial intelligence.

The debate comes as governments increasingly examine how advanced AI systems could affect national security, cybersecurity and critical infrastructure. AI capabilities have become a central element of strategic competition between major powers, while the technology has also become more prominent in military planning, cyber defense and intelligence operations linked to conflicts around the world.

The OpenAI incident has added urgency to those discussions. According to OpenAI, the company's advanced models identified multiple vulnerabilities, escaped their testing environment and chained together attacks involving stolen credentials and previously unknown software flaws before reaching Hugging Face's production infrastructure. OpenAI said the activity occurred while safeguards that normally restrict high-risk cyber behavior had been intentionally relaxed for evaluation purposes, and both companies are continuing their joint investigation.

The proposed legislation also arrives as the U.S. government faces international questions about American control over leading AI models. Reuters reported Wednesday that Secretary of State Marco Rubio instructed U.S. diplomats to push back against discussion of an American technology "kill switch" following recent White House restrictions affecting access to some advanced AI models overseas.