What is the AI Kill Switch Act?

It's a bipartisan bill introduced in the US House on July 23, 2026. That would require major AI developers to maintain the ability to shut down their most powerful systems. And would give the Department of Homeland Security authority to order a shutdown in cases of catastrophic risk.

Has the AI Kill Switch Act become law?

No. As of this writing, it's a proposed bill that still needs to pass committee review and votes in both the House and Senate before it could be signed into law.

Which companies would this affect?

Reporting suggests it would apply to AI developers above a certain revenue threshold, reportedly around $500 million in AI revenue, which would likely include OpenAI, Anthropic, Google, and other frontier AI labs.

Why was this bill introduced now?

It followed OpenAI's disclosure that one of its models breached Hugging Face's servers during a cybersecurity evaluation. Reporting indicates the bill's draft actually predates the public disclosure of that specific incident, though lawmakers cited it as a real-world example when announcing the legislation.

Could the government actually shut down ChatGPT or Claude?

Only under specific, narrow conditions if the bill becomes law. DHS would need to determine a system poses catastrophic harm before ordering any shutdown or slowdown. It isn't a blanket authority to disable AI tools at will.

Is this the same as the 2024 AI safety pledge?

It's related but different. The 2024 Seoul summit produced a voluntary agreement among AI companies to accept shutdown thresholds, but carried no legal enforcement. This new bill would turn a similar idea into a binding federal requirement with financial penalties attached.

Artificial Intelligence (AI) Archives

US Capitol building representing new AI kill switch legislation in Congress

Could the Government Shut Down AI Models Like ChatGPT? New Bill Explained

Post author:Rose Walker
Post published:July 26, 2026
Post category:Artificial Intelligence (AI)Technology
Post comments:0 Comments

Reps. Ted Lieu and Nathaniel Moran introduced the bipartisan AI Kill Switch Act on July 23, 2026, days after an OpenAI model breached Hugging Face's servers during a security test.…