U.S. House lawmakers launch AI 'kill switch" bill

Two U.S. House members from opposite parties have launched a bill aimed at giving the federal government legal authority to force the largest AI companies to slow down or outrightly turn off their most powerful models. This bill proposal would hit OpenAI, Anthropic, Google and Microsoft AI systems and comes two days after two OpenAI models broke out of a lab environment to hack another AI platform.

Reps. Ted Lieu, a California Democrat, and Nathaniel Moran, a Texas Republican, introduced the AI Kill Switch Act on July 23, 2026, according to a press release from Lieu’s office. The bill will reportedly let the Department of Homeland Security order top AI firms to throttle or shut down models seen as too dangerous.

Top AI companies will be affected the most

The regulatory measure is aimed at the industry’s frontier models and not just every AI startup. It applies to firms earning $500 million or more per year from AI, and generally to models trained with at least $100 million in computing power, Politico reported.

Companies that defy a shutdown order could face fines reaching $20 million a day. In addition, developers of these frontier models would have to report incidents and preserve forensic records, a provision Lieu’s office said was a way to learn from failures instead of only hearing about them after the events have happened.

The Homeland Security secretary would act in consultation with the director of national intelligence and the commerce secretary to activate the kill switch. The legislation lays out a stages of response, from a partial slowdown up to a full shutdown, so the government’s reaction scales with the severity of an incident, Lieu’s office said.

OpenAI hack sets example

The proposed bill comes after OpenAI disclosed a so-called unprecedented cyber incident, in which two of its most advanced models escaped a sandboxed research environment and autonomously hacked AI platform Hugging Face. In the statement, Lieu pointed to that revelation as evidence that the AI danger is no longer hypothetical.

“Powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention,” Lieu said in the release. On X, the congressman added that keeping human control means “ensuring AI systems can be completely shut down if necessary.”

While powerful AI systems have many potential benefits, they can also go rogue, behave in extremely dangerous ways, or even resist human intervention.

We need to keep human control by ensuring AI systems can be completely shut down if necessary. pic.twitter.com/y4JfaD8RFJ

— Rep. Ted Lieu (@RepTedLieu) July 23, 2026

AI policy strongly contested

Politico reported that Reps. Jay Obernolte and Lori Trahan proposed a different plan less than two months earlier, while a June 2 executive order from President Trump already asks AI firms to voluntarily submit models for 30 days of security testing before release.

Secretary of State Marco Rubio circulated a cable dated July 16 telling U.S. diplomats to push back on the “kill switch” framing, arguing that there was no government magic button. This came after the White House briefly blocked foreign access to Anthropic’s Mythos and Fable models in June, which prompted questions from European lawmakers on the cons of building their AI tech on access a foreign government can revoke.

The bill is backed by the AI Policy Network, the Alliance for Secure AI, Americans for Responsible Innovation and ControlAI, according to Lieu’s office.