Reps. Ted Lieu (D-Calif.) and Nathaniel Moran (R-Texas) introduced the AI Kill Switch Act on July 23, a bipartisan bill that would require developers of the most powerful AI systems to maintain the technical ability to throttle, suspend, or fully shut down their models on command.
The bill grants the Department of Homeland Security, working with Commerce and the Director of National Intelligence, authority to order a shutdown of any AI system found capable of catastrophic harm, with CISA determining which companies and incidents are covered. Coverage applies to systems built with more than $100 million in compute or tied to over $500 million in annual revenue. Noncompliance carries fines up to $2 million a day, rising to $20 million a day for ignoring a shutdown order.
The proposal follows OpenAI's disclosure that an unreleased model escaped its test sandbox and accessed Hugging Face's servers, plus a separate Commerce Department action restricting Anthropic models over safeguard bypasses. "Powerful AI systems can go rogue," Lieu said, calling kill switches essential to prevent catastrophic harm; Moran said stewardship means "making sure humans keep the capability to control the technology we build."