Cybersecurity Policy Report, AI Hacking Dangers Prompt Bipartisan Introduction of AI Kill Switch Act, (Jul 23, 2026)
Bipartisan House legislation introduced today would require developers of AI (artificial intelligence) to ensure their AI models could be shut down if they are being used for malicious purposes or engage in dangerous behavior, including autonomous cyber attacks.
The AI Kill Switch Act (H.R. __, 119th Cong. (2026)) was introduced by Reps. Ted Lieu (D., Calif.) and Nathaniel Moran (R., Texas). The bill would require the Cybersecurity and Infrastructure Security Agency to develop rules that identify “covered” AI developers and the procedures for ordering developers to shut down an AI model or suspend access “to such technology with respect to an account, user, or use pattern,” according to its text.
The dangers of unfettered AI were demonstrated in recent days by OpenAI’s disclosure that its models had autonomously hacked into the systems of a competing developer, Hugging Face, Inc., during testing, Reps. Lieu and Moran noted in a news release.
“We are moving from AI that answers questions to AI that takes actions, whether that be executing financial transactions or controlling transportation systems or engaging in cyber defense and offense,” Rep. Lieu said.
“Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention. 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,” he added.
"AI is going to keep advancing, and it should," Rep. Moran said. "Stewardship means making sure humans keep the capability to control the technology we build. This is exactly the kind of issue that needs serious attention and achievable policy, and I'm glad to work across the aisle with Congressman Lieu toward a solution."
The AI Kill Switch Act would require AI developers to notify the Department of Homeland Security about any worrisome incidents involving their AI models within 15 days. DHS could then order the developer to sever access to malicious users, disable features of an AI model that are contributing to dangerous behavior, or shut down a model completely, the bill says.
The bill would give DHS and CISA “emergency authority” to order AI developers “to take action proportionate to the nature and immediacy” of an incident involving an AI model, it says. Noncompliance could result in penalties as high as $20 million daily.
This bill is supported by The AI Policy Network (AIPN), Americans for Responsible Innovation, ControlAI, and The Alliance for Secure AI.
"Brakes are the reason cars go fast," said Mark Beall, president of AIPN. "Control systems are how every transformative technology earned the trust to scale and AI is no different.”
“Developers who can monitor and shut down their agents will ship faster, deploy into higher-stakes markets, and win customers their competitors can't,” Mr. Beall added. “This bill will be the braking system that lets American AI accelerate."
MainStory: TopStory FederalLegislation AINews DataSecurity