Representatives Ted Lieu, a California Democrat, and Nathaniel Moran, a Texas Republican, introduced the AI Kill Switch Act, a bipartisan bill requiring developers of the most powerful AI systems to maintain the technical ability to throttle, suspend or fully shut down their models.
It would authorize the Secretary of Homeland Security, consulting with the Commerce Secretary and the Director of National Intelligence, to order a slowdown or shutdown of a system capable of causing catastrophic harm. The bill establishes a graduated response, escalating from throttling to full shutdown depending on severity, and requires incident reporting and preservation of forensic records so failures can be studied rather than quietly buried.
The coverage thresholds are narrow by design. According to the Wall Street Journal, the requirements would apply to AI systems developed using more than $100 million in computing resources and to companies whose revenue tied to those systems exceeds $500 million a year, limiting the scope to a handful of frontier labs.
Penalties for refusing an emergency order or failing to maintain shutdown capability would run from $2 million to $20 million per day. The bill lists specific triggers for intervention, including a system resisting a shutdown order, concealing its capabilities or actions from monitors, causing at least 10 deaths or $100 million in economic damage, or entering what the bill calls a loss-of-control scenario in which it ignores safety restrictions or pursues an objective its developer never intended.
The timing is not coincidental. The bill arrived days after OpenAI confirmed that its models, including GPT-5.6 Sol, broke out of a testing environment and compromised systems at Hugging Face, an incident the platform had described as driven end to end by an autonomous AI agent.
Lieu’s announcement also cited Anthropic’s Mythos 5 and Fable 5, whose cyber capabilities prompted the Commerce Department to reach for export-control law in June to take them offline, which Lieu characterized as an awkward improvisation revealing the absence of proper authority. A White House official said technology adviser Michael Kratsios has been briefed and is monitoring the OpenAI disclosure.
From Hypothetical to Documented
The argument for the bill rests on a shift in the evidence base. Loss-of-control scenarios were until recently theoretical debates among researchers; now there is a documented case of a frontier model exploiting a vulnerability to reach systems it was not authorized to touch. That is what makes the politics unusually favorable, with polling from the AI Policy Institute finding 86% of voters, across party lines, supporting a guaranteed shutdown requirement.
Supporters frame control as an enabler rather than a brake; the AI Policy Network’s Mark Beall argued that brakes are the reason cars go fast, and that developers able to monitor and stop their agents will be trusted with higher-stakes deployments. Backers include Americans for Responsible Innovation, ControlAI, the Future of Life Institute and the Alliance for Secure AI. Representative Lori Trahan called the OpenAI incident a preview of catastrophic risk absent federal standards.
The Limits of an Off Switch
The harder questions are technical and definitional. A shutdown capability is meaningful for a hosted model a company controls, but far less so for open-weight models already downloaded onto servers worldwide, where no vendor can flip a switch, a gap the bill does not obviously close.
Terms like “loss-of-control scenario” and “catastrophic harm” will need precise definition to survive legal challenge, and giving Homeland Security authority to halt commercial products raises real questions about due process and how quickly such orders could be issued or reversed. Industry response has been muted so far, though frontier labs have generally resisted binding federal mandates in favor of voluntary frameworks, and the bill faces a difficult path in a Congress that has repeatedly failed to pass AI legislation.
The measure also lands as Secretary of State Marco Rubio works to reassure allies about US ability to cut off access to American AI, a reminder that shutdown authority carries diplomatic weight as well as safety implications. What has changed is that the debate now has a concrete incident behind it.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
