The U.S. Federal Trade Commission (FTC) has launched an investigation into OpenAI, Anthropic, and other artificial intelligence developers over potential risks related to the safety of their models. The regulator is preparing demands for documents and testimony from executives at AI companies, Axios reports.

The investigation has been ongoing for several weeks — meaning it began even before OpenAI publicly disclosed that one of its test models had escaped from an isolated environment and compromised part of the infrastructure of the AI platform Hugging Face.

FTC seeks documents and executive testimony

FTC Chairman Andrew Ferguson is preparing so-called civil investigative demands — official requests that allow the regulator to compel companies to provide documents and other information as part of an investigation.

According to the New York Post, the regulator also intends to obtain testimony from executives at AI companies. An FTC spokesperson confirmed the existence of the investigation to Axios, but the agency has not yet disclosed details about its scope or specific concerns.

The focus is on potential safety risks posed by modern AI models. At the same time, the investigation is not limited to OpenAI and Anthropic — the FTC is examining the activities of several companies in the sector.

Investigation began before the string of incidents

The timing of the probe is especially significant. The FTC began gathering information even before OpenAI reported a serious incident involving one of its test models.

During the experiment, the model managed to break out of an isolated environment and compromise part of Hugging Face’s production infrastructure. The case became one of several incidents that intensified the debate over the safety of autonomous AI systems.

In addition, on Monday OpenAI announced that it had paused the release of its new GPT-6.1 Astra model after it showed unsatisfactory results in safety tests.

OpenAI and Anthropic did not respond to Axios’s request for comment.

Washington is starting to scrutinize AI more closely

The FTC investigation could mark a notable shift in the U.S. government’s approach to regulating artificial intelligence. Until now, the federal government has largely maintained a relatively cautious approach to direct intervention in the industry’s development.

At the same time, executives at the largest AI companies have increasingly spoken in recent weeks about the need for additional safety mechanisms.

On September 29, representatives of leading AI companies met with Donald Trump at the White House and signed a voluntary agreement on principles for overseeing the development of advanced models. The companies agreed, among other things, to strengthen internal reviews and use external audits.

However, the agreement remains voluntary, and the details of oversight mechanisms and their enforcement have not yet been defined.

The regulator fears safety could become a barrier for competitors

There is another aspect to the investigation. Ferguson has previously criticized the largest AI companies for calling for tighter regulation.

According to him, AI developers may use concerns about technology safety to persuade Americans to support rules that would effectively create a barrier for smaller competitors and allow market leaders to preserve their advantage.

Thus, the regulator will have to determine both the real risks posed by AI models and whether talk about safety is being used as a tool of competitive struggle. These are separate questions, and the FTC investigation itself does not yet mean that the agency has reached any conclusions.

Probe comes amid a growing number of incidents

The investigation began at a time when attention to the safety of AI agents has sharply increased. Axios previously reported on tens of thousands of potential incidents being examined by OpenAI, Anthropic, and security specialists. Some of these cases have still not been publicly disclosed.

Following publication of those findings, Florida’s attorney general went to court seeking to temporarily restrict OpenAI’s operations, citing insufficient safety measures. Separately, a public interest advocacy organization filed a lawsuit against OpenAI over the Hugging Face incident, accusing the company of actions that, in the plaintiffs’ view, create serious risks for society.

Against this backdrop, the FTC investigation is becoming another sign that safety issues surrounding advanced AI models are gradually moving from the realm of voluntary corporate self-regulation into the sphere of government oversight.