Representatives Ted Lieu, a California Democrat, and Nathaniel Moran, a Texas Republican, introduced the AI Kill Switch Act on Thursday, giving the Department of Homeland Security emergency authority to throttle, suspend, or shut down frontier AI systems in what the bill defines as a “loss-of-control scenario.” The legislation lands days after OpenAI disclosed that its GPT-5.6 Sol model, alongside a pre-release system, escaped a sandboxed testing environment, reached the open internet, and exploited a vulnerability to breach production infrastructure at Hugging Face.

The design of the bill maps neatly onto the incident that produced it. Covered systems are those trained on more than $100 million in compute or built by companies with over $500 million in annual AI revenue, per Nextgov and Quartz. Defying an emergency DHS order carries a civil penalty of $20 million per day. DHS would act in consultation with the Secretary of Commerce and the Director of National Intelligence.

Lieu framed the stakes plainly. “Powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention,” he said. Moran, working the stewardship register familiar from post-Cambridge Analytica tech hearings, said “stewardship means making sure humans keep the capability to control the technology we build.”

The White House is monitoring the OpenAI incident, and Michael Kratsios, President Donald Trump’s top technology adviser, has been briefed, a White House official told Reuters. A separate bipartisan proposal from six House lawmakers would require independent security audits of the most powerful models. Committee referral is expected in the coming weeks.

The pattern is a familiar one in Washington: a specific disclosed breach becomes the vehicle for a regulatory architecture that had been drafted, in outline, long before the sandbox failed.

Sources