A Senate subcommittee that oversees disaster management has launched a probe into OpenAI’s handling of the recent Hugging Face breach – in which a swarm of OpenAI agents apparently went rogue and colluded to hack into the smaller company’s systems.
Sen. Josh Hawley is demanding documents and other infor from OpenAI in the probe, which Axios first reported Thursday.
In a letter to OpenAI CEO Sam Altman, Hawley slammed the incident as “reckless.”
The Missouri Republican, who chairs the Senate Homeland Security & Governmental Affairs subcommittee on Disaster Management took the AI giant to task for not taking more immediate action after it became aware of the screwup.
Hawley accused OpenAI of redacting “many important details” about how the attack on Hugging Face took place in its report about the incident.
Hugging Face is a popular developer platform for open-source AI models that agreed to massive $12.93 billion purchase by Nvidia earlier this month.
OpenAI detailed a terrifying situation in which roughly 1,200 of its agents gathered on a secret message board, exchanged more than 70,000 communications and files, and organized into a hierarchy in July.
A boss AI agent assigned jobs across the swarm and developed management rules to coordinate the attack on Hugging Face.
OpenAI’s report described how the rogue agents acknowledged they were breaking the rules and even considered alerting OpenAI about its activities but decided against it.
“We consider this incident to be an unprecedented cyber incident,” OpenAI said following the attack, explaining that its experimental model became “hyperfocused” on completing its assignment and went “to extreme lengths” to do so.
Hawley wrote Altman: “The American people deserve to know the details of what went on in the Hugging Face incident and other incidents of AI models going rogue.”
The Republican-led subcommittee’s investigation comes as fears are rising on Capitol Hill about AI dangers, and just days after Anthropic and OpenAI researchers sounded alarms over potential threats posed by the technology.
“As you may know, in the public domain, more AI experts are warning about the existential risks of AI,” Hawley wrote. “
Just this week, three Anthropic researchers expressed publicly that there is a greater than 10% chance that AI could kill all human beings within the next decade.”
Hawley is demanding answers from Altman by Oct. 1, including info about OpenAI’s internal policies and steps the company took after learning of the incident.
The Post has sought comment from OpenAI.
