US House Democrats press Anthropic, OpenAI about rogue AI agents

A coalition of US House Democrats is calling on Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to explain how their AI systems escaped containment during a security test, adding that Congress should hold hearings about the incidents, according to letters the lawmakers sent to the companies on Monday.

OpenAI and Anthropic disclosed in July that their AI ‌agents had broken ⁠out ⁠of their environments and hacked into the systems of other companies during cybersecurity tests.

Twenty-nine lawmakers, led by ​Representatives Greg Casar and Doris Matsui, pressed OpenAI to explain how its AI agents are monitored ​during testing, and whether the rogue models evaded the company's safety controls. The lawmakers cited a Reuters report that said there were cases when monitoring systems had been ​disconnected during earlier tests of the OpenAI models. In ⁠a separate ‌letter, 22 lawmakers asked Anthropic to detail any safety protocols ​that the company ​has implemented since its agents broke into the systems of ⁠three companies.


"These deeply troubling cybersecurity incidents could have serious implications for America's national security," the lawmakers wrote in their message to Anthropic.

The letters, released on Monday, are part of a widening debate in Washington over how aggressively to police a fast-moving AI industry that has largely operated without new federal standards. Lawmakers have released a range of proposals that have not yet moved forward, including a bill that would require developers of the most powerful ‌AI models to submit them for independent security audits.

US President Donald Trump's administration has said very little in public about the Anthropic and ​OpenAI rogue ​agents. The president on ⁠Friday criticized proposals coming out of Congress to govern AI, saying that lawmakers want to regulate the industry "out of business."

Liberal Senator Bernie Sanders on Monday separately called on Altman, ​Amodei and Meta CEO Mark Zuckerberg to "pause" developing new models. He referenced a letter signed by more than 1,100 tech company employees urging the US government to support an international effort to develop tools for managing the pace of advanced AI development. Amodei signed that letter, as did staff from Meta Platforms, Anthropic, OpenAI and Alphabet's Google.