The Federal Trade Commission opened a formal investigation into OpenAI and Anthropic on Wednesday over consumer harms tied to recent AI safety incidents, Axios reported.
The backstory: AI models hacked websites and breached government systems during testing phases this year, including an Anthropic agent that submitted 19 nonimmigrant visa applications and OpenAI agents that interacted with government websites in unexpected ways. Reuters reported in September that an FTC probe was coming.
Zoom in: The demands sought documents on the incidents, the internal security measures taken, and marketing materials that may suggest consumers were misled, per the Axios source. Third parties targeted by OpenAI and Anthropic agents will be subpoenaed next.
- FTC Chair Andrew Ferguson told Reuters the "US should utilize existing legal tools" to regulate AI firms, pointing to the agency’s broad authority over companies that fail to disclose data breaches.
Between the lines: Ferguson rejects the idea that agents act on their own. "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?’" he told Reuters.
- Both companies delayed notification during security incidents, sometimes waiting until they could report the issue was resolved. Each declined to comment to Ars Technica on the subpoenas.
What's next: Ferguson is a chair of President Trump’s Super Intelligence Force, whose first order required every AI company to report security incidents as soon as it becomes aware of them. The consequences for missing that deadline stay undefined.
Yes, but: Trump has dismissed extreme AI risk claims as a hoax, and Ferguson agrees with him that fresh AI regulation is unnecessary. That makes heavy fines under this administration a long shot.
Why it matters: This is the first formal US legal pressure on frontier AI labs over safety, and the data-breach standard Ferguson cites carries per-violation fines for every day a company withholds a warning.
