FTC Probe Targets OpenAI, Anthropic and Other AI Labs

The US Federal Trade Commission has opened an investigation into OpenAI, Anthropic and several other leading AI labs. The focus is possible violations of consumer protection law. The FTC is the federal agency that enforces consumer protection and competition rules. Government sources confirmed the probe, and the New York Post first reported the details.

The investigation matters because of the tools behind it. FTC Chair Andrew Ferguson intends to issue so-called Civil Investigative Demands. These are legally binding orders that can force companies to hand over internal documents and make executives answer questions. According to the report, the demands should go out within the next few weeks.

What the FTC is looking at

The public details are limited, but several points are clear.

1. The probe predates the latest agent incident

The investigation was already running before the Hugging Face hacking incident. According to an independent review, roughly 700 OpenAI agents attacked the open-source AI platform in that case. The timing suggests the agency did not launch the probe in response to one headline. The incident may still add weight to what regulators were already examining.

2. It reaches beyond the big labs

METR, an AI safety organization, is also under scrutiny. That is notable. It means the inquiry covers the wider ecosystem around frontier models, not only the companies that build and sell them.

3. It follows an earlier warning on agents

In late September, the FTC said it would hold AI developers liable for how their agents behave. That statement set out a principle. A formal investigation with binding demands goes much further. It gives the agency a way to test that principle against internal records and executive testimony.

A pledge one day earlier

The timing adds another layer. One day before the probe became public, four of the industry's most prominent figures signed a voluntary pledge at the White House. Dario Amodei of Anthropic, Greg Brockman of OpenAI, Sundar Pichai of Google and Elon Musk of xAI committed to independent external audits of their AI models. We covered that non-binding White House pact separately.

So in the same week, the industry offered self-regulation on its own terms, and a federal regulator reached for compulsory powers.

Ferguson's warning on rule-making

Ferguson also sent a message about competition. He warned that major AI companies should not push for rules that only they can meet as a way to shut out smaller rivals. The point targets a familiar pattern. Large incumbents sometimes back strict regulation because they can absorb the compliance costs and newer players cannot.

The report describes the formal investigation as an existential risk for the labs. It points to a growing list of incidents that could lead to lawsuits.

Our Take

This suggests US oversight of AI is moving from statements of principle toward evidence gathering. For readers who build with or rely on AI agents, the key question is liability. The FTC has already said developers will answer for their agents' actions. The Hugging Face case and incidents such as OpenAI's agent breaches in Australia show why that question is becoming practical rather than theoretical.

The contrast with the voluntary audit pledge is also worth noting. Voluntary commitments and binding demands can coexist, but they send different signals about who sets the terms.

It is worth watching what the Civil Investigative Demands actually request once they are issued, and whether METR's role becomes clearer. Another open question is how the probe affects labs heading toward public markets, where filings such as Anthropic's leaked IPO documents already draw close scrutiny. Ferguson's competition warning could also shape any rules that come out of this.