WASHINGTON — Hudson Valley Rep. Mike Lawler and New Jersey Rep. Josh Gottheimer have rolled out a new bill aimed at reining in artificial intelligence agents and preventing them from going rogue.
The “Stop Rogue AI Act” comes on the heels of recent resignations by Anthropic safety researchers and revelations from OpenAI about the Hugging Face incident in which its AI agents hacked into another company.
“It’s not a movie plot, right?” Gottheimer (D-NJ) said. “You think it sounds like a Terminator sequel, but it’s unfortunately actually happened, and you keep hearing about more incidents that are being reported beyond Hugging Face that they have found in the pre-deployment phase of these models.”
AI companies such as OpenAI and Anthropic, which have been bleeding through billions of dollars worth of cash at a fervent clip and struggled to get a return on investment, have alarmed the public with their recent disclosures about AI faux pas.
After Anthropic safety researchers Joe Benton and Jacob Coxon quit in protest earlier this month, CEO Dario Amodei, who had tried to build a reputation of being cautious, called for a slowdown in AI development.
Amodei was backed up by OpenAI CEO Sam Altman and xAI boss Elon Musk, among others, in that call. President Trump has blasted AI doom predictions as “things that won’t happen” and insisted that the US remain the leader of the nascent technology.
“My biggest concern with AI from the beginning has been the use of AI by foreign bad actors like China, like Russia, like Iran to undermine our national security and undermine our financial markets, our institutions, and our capitalist system,” Lawler told The Post.
“We know China seeks to do that every day,” he added. “There’s no doubt in my mind that they will try to use AI to undermine and attack our systems, which is why the cybersecurity component is so critical, and it’s why understanding what is happening with these rogue AI agents is so critical.”
The “Stop Rogue AI Act” instructs the National Institute of Standards and Technology (NIST) to set up guidelines and standards for controlling AI agents. It calls for monitoring the AI agents “in real time” and ensuring that tech companies are able to figure out what they can do.
“It is about making sure that we can safely deploy AI,” Lawler said. “If the United States taps the brakes while others floor it, we do not get safer AI. We get someone else’s AI. We are not going to regulate ourselves into second place.”
Both Lawler and Gottheimer have teamed up on multiple bipartisan bills, such as the National Security Commission Quantum Computing Act, Advancing Quantum Leadership Act, and AI Workforce Training Act.
Their latest bill comes as the House has opted to break early for recess to give members more time to campaign for the midterm elections.
