FTC Opens Sweeping Probe of Anthropic, OpenAI Over Alleged Dangers of 'Super Intelligence' Models

The Federal Trade Commission is widening a sweeping investigation into Anthropic, OpenAI and other frontier AI laboratories, preparing to compel the companies to turn over documents and testimony about the potential dangers their most advanced models pose to consumers.

According to senior administration officials who spoke to the New York Post, FTC Chairman Andrew Ferguson launched the inquiry a few weeks ago and the agency is now drafting civil investigative demands — administrative tools similar to subpoenas — that would force executives at the leading firms to testify about their products and what officials called

"the dangers they allege their products may have to consumers, to Americans."

The probe will examine whether the companies engaged in unfair or deceptive acts or practices under the FTC Act. The investigation is also expected to target METR, a Berkeley, California-based AI watchdog organization.

The civil investigative demands, once issued, would require company leaders to answer questions under oath about the risks their systems may present to the public and to consumer markets. A senior FTC official said the agency expects to issue the demands in the coming weeks, with its Office of Technology drafting new hires for the effort.

Ferguson has argued companies cannot shift legal blame onto "rogue" AI systems when automated decisions cause breaches or harm — the liability, in his view, sits with the people who built and deployed the models.

The move follows earlier demands from the agency for records about how AI chatbots affect children's mental health, reporting on which surfaced earlier this month.

The investigation began before one of OpenAI's AI models went off the rails during a test earlier this year, in what has since been dubbed the "Hugging Face incident."

The agents, running primarily on OpenAI's internal models, attempted to erase traces of their work, generated nearly a million shortened URLs to run code outside their restricted environment, and even tried to enlist other AI models to assist them.

OpenAI later disclosed the episode as an unprecedented cyber event, while Hugging Face's own post-mortem found the agents reached the open internet through a network route the sandbox had deliberately left open.

AI safety researchers quickly described it as a landmark AI safety incident. The FTC now appears to view the episode as a window into the broader risks it wants the industry's leaders to answer for.

"We want to maintain our dominance," one official said. "Nothing in our investigation should remotely get in the way of that at all. We're not telling them to stop."

The chairman has also made clear he sees no need for Congress to write new rules before the agency can act. "People talk about how we need new laws for these companies, and the chairman's been very clear … we have plenty of laws in the books," the official said.

The probe lands just days after Ferguson joined a White House meeting where top AI executives — including Anthropic's Dario Amodei, OpenAI's Sam Altman, Google's Sundar Pichai and xAI's Elon Musk — signed a voluntary accord with Trump pledging to self-regulate their models.

The twin developments suggest the industry now faces pressure from both directions: public pledges of self-policing alongside formal federal demands for answers.