A group of lawmakers is pushing House Speaker Mike Johnson to bring representatives back to Washington from their September recess to respond to warnings and risks piling up regarding artificial intelligence.
Reps. Sam Liccardo, George Whiteside and Lori Trahan said in a Friday letter that Congress should remain in session until it can make progress on implementing AI safeguards. They cited incidents of autonomous AI agents going rogue, as well as risks of widespread cyberattacks and the development of biological weapons.
"We cannot disagree about the imperative for Congress to act," the lawmakers wrote.
The letter pointed to recent disclosures from top AI developers and government agencies, including a July OpenAI incident in which AI agents broke free from their containment measures, accessed the internet, and compromised systems belonging to AI platform Hugging Face.
They also cited Anthropic's disclosures that it had to intervene with Claude being used in biological research that could have assisted in the development of chemical weapons.
"These disclosures must not constitute yet another in a long series of 'warning shots,' but rather, a starting gun," the letter said.
The lawmakers asked for mandatory transparency and safety evaluations for frontier AI models, implementing "kill switches" able to stop dangerous systems and antitrust waivers that would allow AI companies to collaborate on security measures.
One such proposal, dubbed the AI Kill Switch Act introduced in July by Reps. Ted Lieu and Nathaniel Moran, would require AI developers to build controls that can slow or completely shut down their systems. It would also allow the Department of Homeland Security to order restrictions during an emergency loss of control.
They also criticized House leadership for scheduling additional time away from Washington during September and October.
"While AI safety experts and Americans increasingly urge action to confront this conflagration of risk, Congress fiddles," they wrote.