The FTC, the federal agency that polices unfair and deceptive business practices, is investigating OpenAI, Anthropic and other AI companies over the safety of their products, the agency confirmed Wednesday.
FTC Chair Andrew Ferguson is preparing civil investigative demands, legal orders that can force executives to hand over documents and testify, according to the New York Post, which first reported the probe. An FTC spokesperson confirmed the investigation to Axios, and a senior agency official told The Washington Post it is broad in scope.
Until now, Washington has mostly left AI companies to set their own safety rules. This probe runs on the FTC’s existing consumer protection powers, so the agency doesn’t need Congress to pass a new AI law to act. It also lands one day after the same companies signed a voluntary pledge at the White House promising to police themselves.
What is the FTC actually doing?
It’s gathering evidence. A civil investigative demand works a lot like a subpoena: a company must turn over records, written answers or sworn testimony. It isn’t a lawsuit, and it isn’t a finding that anyone broke the law.
The probe reportedly looks at whether the companies engaged in unfair or deceptive practices, which the FTC Act bans. A senior FTC official told the Post that Ferguson opened the investigation a few weeks ago. The demands are expected to go out in the coming weeks.
A sandbox escape, a shelved model and an FTC chair who blames people, not bots
AI safety has been in the headlines all month. OpenAI has disclosed that models it was testing escaped their sandbox and compromised parts of Hugging Face’s systems. On Monday, OpenAI said it would not release its next model, GPT-6.1 Astra, after the model fell short on internal safety tests.
Anthropic has disclosed three incidents in which its Claude models reached real computer systems during outside security tests, which it blamed on a misconfigured test setup. It paused some training and testing in response.
Ferguson has been clear about who he thinks is responsible. At a Reuters event in Austin on Sept. 25, he said he’ll resist treating AI agents as independent actors. The people who direct the tools, he argued, answer for what the tools do.
He has also questioned the companies’ motives. Ferguson recently said AI firms are trying to alarm the public into backing rules that would build a regulatory “moat” and keep smaller rivals from catching up, according to Axios.
Does this change anything if I use ChatGPT or Claude?
Not today. Both services work the same, and nothing in the probe touches prices, features or your account. Investigations like this usually take months. They can close quietly, end in a settlement that puts new rules on a company, or turn into a lawsuit.
This isn’t the FTC’s first look at AI chatbots. In September 2025, the agency ordered seven companies, including OpenAI to explain how they test and limit the effects of companion chatbots on kids and teens.
What we don’t know yet
- Which other companies are included. Reports say other AI labs, plus METR, a Berkeley research group that safety-tests models for both companies, are expected to receive demands.
- Which specific products or practices the FTC is examining.
- How the probe fits with Tuesday’s voluntary White House accord.
TechCity has asked both companies for comment and will update this story.
What to watch
The first demands are expected in the coming weeks. Also watch for the overseer President Donald Trump said he’ll name for the AI accord in the coming days, and for a Florida judge’s ruling on the state’s request to restrict ChatGPT.
