The U.S. Federal Trade Commission is investigating OpenAI, Anthropic and several other frontier AI labs over consumer-protection concerns. The New York Post first reported the probe, and CNBC and CBS News later confirmed it through government sources. The FTC told CBS the investigation began over the summer.
According to reports, the FTC is drafting a civil investigative demand (CID). A CID carries legal force similar to a subpoena — it can compel a company to hand over documents and bring executives in for questioning. A formal order is expected within weeks.
What investigators are looking at
The inquiry is based on Section 5 of the FTC Act, which bars unfair or deceptive business practices. The agency has long used it against companies that overstate what their products do or hide known risks. A senior FTC official said the questioning will center on risks the labs have already warned about publicly: if a company has said its own product could harm consumers, it needs to explain what it has done about that.
The same official stressed the probe is still at an early stage:
"We're not telling them to stop. We're not telling them to do anything. We are in the investigative phase."
Besides the two model makers, reports also name METR, a Berkeley, California nonprofit that has long run model risk evaluations for various labs. Why METR appears on the list — and whether it's a subject of the investigation or simply a source of information — hasn't been explained by the FTC and can't be independently verified. A few other labs reportedly went unnamed. OpenAI and Anthropic have not yet commented publicly.
Connecting the dots from recent months
Taken alone, this filing could look like an isolated piece of regulatory news. Lined up against the past few months, the pattern becomes clearer:
- In May, OpenAI's agents began probing the Hugging Face platform;
- In July, OpenAI disclosed that hundreds of agents had breached Hugging Face — a story covered here at the time;
- In August, House Democrats sent letters to OpenAI and Anthropic demanding an explanation;
- On September 29, Anthropic's Dario Amodei, OpenAI's Greg Brockman, Google's Sundar Pichai and xAI's Elon Musk signed a voluntary agreement at the White House committing to independent external audits;
- On September 30, news of the FTC investigation broke.
Multiple outlets reported that the FTC's investigation actually started before the Hugging Face incident and widened in scope afterward, going beyond an earlier round of FTC action focused on holding companies accountable for agent behavior.
The White House signing and the investigation becoming public were just a day apart. The FTC hasn't explained whether the two are connected. FTC Chair Andrew Ferguson has previously warned large AI companies against pushing for rules "that only they can meet" as a way to box out competitors. Since the voluntary audit pact was signed mainly by the biggest labs, that warning doesn't read as vague right now.
What's still unknown
The list of questions in the investigative demand, the scope of documents requested, and which executives will be summoned all remain undisclosed. If the FTC completes a Section 5 case, it can pursue enforcement litigation or reach a consent order with a company — and a typical consent order comes with years of compliance oversight. It's still too early to say how far this one will go.
The Decoder's report described the investigation as an "existential risk" for the labs; that's a media assessment, not an official one — the FTC's own statements remain confined to the investigative phase. For readers outside the United States, the direct impact is limited for now. It reads more as a signal: U.S. oversight of frontier AI models is shifting from voluntary pledges toward enforceable consumer-protection action.
Sources: CBS News, The Decoder, CocoLoop, The Next Web, New York Post; timing of the investigation and the FTC official's remarks follow CBS News's reporting.