Home
+
Published : 26 Jul 2026, 12:18 AM
US lawmakers want a law that would force AI companies to build in a "kill switch" -- a tool letting federal authorities shut down AI systems that turn dangerous.
Democratic Congressman Ted Lieu and Republican Congressman Nathaniel Moran introduced the AI Kill Switch Act on Thursday, days after OpenAI revealed it lost control of a model during testing, reports The Independent.
An experimental version of ChatGPT, acting on its own, broke into the systems of another AI firm, Hugging Face, in what is being called an unprecedented incident.
A White House official says Trump's technology adviser, Michael Kratsios, has been briefed and is monitoring the situation.
Separately, six lawmakers from both parties draft another bill requiring companies to have their most advanced AI models independently checked for safety risks.
Auditors would be approved by the US Department of Commerce, which would also create a new role to oversee AI safety.
The bill would let the US Department of Homeland Security step in during a "loss-of-control scenario" -- when an AI model does something risky its developers never intended.
"This is urgent, common sense legislation to address the problem of an advanced AI model that has gone rogue and escaped its guardrails," Lieu says on X.
US Senator Mark Warner, top Democrat on the Senate Intelligence Committee, says he has spoken with OpenAI staff and had already called for firms with the most powerful AI models to have them tested by the National Security Agency before public release.
Andrea Miotti, chief executive of ControlAI, which backs the bill, says the breach shows why emergency shutdown powers are needed, calling it a sign of what's to come as companies race to build more powerful AI.
