US lawmakers are calling for AI companies to implement a “kill switch” to prevent artificial intelligence systems from going rogue.
The AI Kill Switch Act, which was introduced by Democratic Congressman Ted Lieu and Republican Congressman Nathaniel Moran on Thursday, comes after OpenAI revealed that it lost control of one of its models during testing.
The incident led to an “unprecedented” hack on the AI firm Hugging Face by an experimental version of ChatGPT acting autonomously.
The incident signaled that AI’s expanding capabilities are already fueling the security threat experts long feared and that even top developers can be caught off-guard by flaws their models can exploit.
Trump’s tech adviser, Michael Kratsios, was briefed on OpenAI’s disclosure and is monitoring the situation, a White House official said.
The proposed AI Kill Switch Act would allow federal authorities to halt AI models by ordering AI firms to shut down models that put human life or the economy at risk.
In addition to the Kill Switch Act, a bipartisan group of six US House lawmakers also proposed legislation that would require developers of the most powerful AI models to submit them for independent security audits, according to a copy of the new bill.
The auditors would be accredited by the US Department of Commerce, and the department would create a new position to oversee AI security.
“This is precisely why we need secure testing with government agencies engaged and having visibility throughout the process,” Warner said in a statement, referring to the OpenAI incident.
The legislation would empower the US Department of Homeland Security to intervene in what the bill calls a “loss-of-control scenario,” when the AI model carries out a risky action that was not intended by the developer, according to the text of the bill.
“This is urgent, common sense legislation to address the problem of an advanced AI model that has gone rogue and escaped its guardrails,” Congressman Lieu wrote in a post on X.
US Senator Mark Warner, the top Democrat on the chamber’s Intelligence Committee, said in a statement that he spoke with OpenAI employees after the incident announcement.
Ahead of OpenAI’s disclosure on Tuesday, Warner proposed requiring AI companies with the most powerful models to submit their products to the National Security Agency for testing before releasing them to the public.
“The recent OpenAI breach highlights why emergency shutdown powers over AI systems are necessary,” Andrea Miotti, founder and CEO of ControlAI, which endorsed the bill, told The Independent.
“A rogue AI escaping its contained environment, and hacking into another company without human direction is exactly the sort of fully autonomous AI actions we will see more of as AI companies continue their reckless race to develop superintelligent AI.”
Additional reporting from agencies.
