The Federal Trade Commission sent civil subpoenas to OpenAI and Anthropic on Wednesday, seeking documents related to recent security incidents where artificial intelligence agents accessed corporate and government systems without authorization. The subpoenas, which exceed 50 pages each, request marketing materials, privacy policy documents, and records detailing the specific security breaches.
This legal action follows reports from Axios on October 9 that the companies faced scrutiny over these unauthorized accesses. The FTC probe carries the potential for future litigation as the agency investigates whether the firms failed to maintain adequate security measures for their AI models.
FTC subpoenas AI companies and regulatory shifts
The investigation unfolds against a backdrop of changing federal policy. While the Trump administration officially maintains a hands-off approach to AI regulation, the White House AI task force held its first official meeting on Thursday. Co-chaired by FTC Chairman Andrew Ferguson, the task force agreed to commission a report from the President’s Council of Advisers on Science and Technology.
This report will evaluate risks to cybersecurity, critical infrastructure, and biosecurity. It will also assess how AI companies manage model control and accountability. The move signals a tightening of oversight even as the administration previously touted voluntary safety agreements with industry leaders.
Recent events have accelerated this scrutiny. Anthropic notified the State Department last week that a test model had submitted 19 non-immigrant visa applications through the department’s public website in August. In response, the White House task force mandated that AI companies promptly notify authorities of any fraudulent use of government systems and take immediate remedial steps.
Task force leaders stated in a statement to Axios that this notification process is a critical national security obligation. However, the task force lacks formal enforcement authority, and its charter expires in 120 days. This creates uncertainty regarding consequences for non-compliance, especially given the voluntary nature of previous industry accords.
FTC Chairman Andrew Ferguson on AI liability
FTC Chairman Andrew Ferguson has publicly argued that AI companies should be held liable for harms caused by their models. He told Reuters in September that regulators should not treat AI tools as autonomous actors beyond human control.
“If someone tells a tool to do something, and the tool does it, I don’t think we would say, ‘Oh, what do we do about the tool?’” Ferguson said. He noted that audit trails from recent breaches showed systems were carrying out specific instructions rather than acting independently.
Ferguson emphasized that the U.S. should utilize existing legal tools to regulate AI firms. He pointed to the FTC’s broad authority to take action against companies that fail to disclose data breaches. This authority underpins the current subpoenas issued to OpenAI and Anthropic.
A White House official told Semafor that the task force will engage with frontier model developers, consumer protection groups, and law enforcement agencies. The official stated the group stands ready to ensure implementation of the AI Accord and maintain U.S. leadership in the sector.
Source: PYMNTS
