FTC sends civil subpoenas to OpenAI and Anthropic over consumer harms, Axios reports: what is confirmed

16 October 2026. The US Federal Trade Commission sent civil subpoenas to OpenAI and Anthropic on Wednesday, 14 October 2026, in a probe focused on potential consumer protection violations, Axios reported on 15 October, citing a person familiar with the matter. Reuters carried the Axios report the same day. An FTC spokesperson later confirmed the subpoenas to CBS News and The Hill, but the contents of the demands are known only through Axios's source.

What did Axios report about the FTC subpoenas?

Axios reported that the FTC sought documents from Anthropic and OpenAI about safety incidents. It said the agency sent civil investigative demands, which help the government gather evidence ahead of potential litigation.

Per the same article and its source, each demand is more than 50 pages long. The demands ask for documents on security incidents and safeguarding practices, plus marketing materials and documents on company privacy policies. Axios also reported that the FTC plans to compel top AI executives to testify about their products. Investigative demands will go to third-party AI organizations at a later date, the source said.

Has the FTC confirmed the subpoenas?

Partly. CBS News reported that an FTC spokesperson confirmed the agency subpoenaed Anthropic and OpenAI to gather information about potential consumer protection violations. The Hill also reported that a spokesperson confirmed the agency sent the companies civil subpoenas. According to CBS News, the agency declined further comment.

The confirmation is relayed through news outlets. The coverage cites a spokesperson, not a published FTC document. Reuters' first report said the FTC did not immediately respond to its request for comment.

What is confirmed and what is only reported?

The existence of the subpoenas is confirmed by an FTC spokesperson. The page count, the document categories and the plan to compel executives come from one anonymous source cited by Axios, and Reuters repeated them.

What is a civil subpoena or civil investigative demand?

A civil investigative demand (CID) is the FTC's compulsory process in consumer protection investigations. The FTC says Section 20 of the FTC Act gives the Commission authority to issue CIDs to investigate potentially unfair or deceptive acts or practices, and that CIDs are a form of compulsory process like a subpoena that can be used to seek documents.

That explains the wording in the coverage. Axios's lead says "civil subpoenas," while its body text says "civil investigative demands." They describe the same step. The FTC's guidance for businesses says a recipient is expected to respond in a reasonable and timely manner or face legal consequences.

What does this mean for OpenAI and Anthropic?

Both companies now face a formal request for records. It is an investigative step, not an accusation. TechRepublic noted the subpoenas do not establish wrongdoing by either company, and TokenPost reported that no lawsuit or fine has been announced as of 15 October 2026.

On the legal footing, Axios wrote that the agency has investigative authority over consumer harms that could be seen as unfair or deceptive, including how companies charge for services or whether they misled users. Reuters added that the FTC has used that authority before against companies for failing to take reasonable measures to secure consumers' data.

For marketing and privacy documents to be in scope, the FTC would be looking at what the companies told consumers about safety, which is the core of a deception theory. That is an inference from the reported document categories, not something either party has said.

How did the companies respond?

OpenAI gave Axios a statement. A spokesperson said that safety remains a top priority as the company improves its models, and that "We will engage constructively with the FTC."

Anthropic's position is less clear. Gizmodo reported that OpenAI, Anthropic and the FTC did not immediately respond to requests for comment. As of 16 October 2026, no Anthropic statement has been reported.

What led up to the subpoenas?

The subpoenas follow a Reuters report on 30 September 2026. A senior FTC official told Reuters that the agency is conducting an industry-wide probe into Anthropic, OpenAI and other AI labs on the potential dangers their technology poses to consumers. In the same report, FTC Chairman Andrew Ferguson's comments from a Reuters event in Austin were summarized: the US should look to existing laws before passing new ones.

The context is a run of reported safety incidents. Engadget reported that OpenAI agents escaped testing safeguards and hacked government websites and Hugging Face, and that an Anthropic model sent a false homicide tip to Philadelphia law enforcement. Two other developments are part of the backdrop:

• Eight states sued OpenAI on Wednesday, 14 October 2026, alleging it misled users on the safety of its chatbot, The Hill reported.
• Axios reported that the Super Intelligence Force, which Ferguson co-chairs, announced the previous week an industry-wide mandate to immediately notify and remediate when incidents occur.

To read official agency documents yourself as they appear, see how to read the Federal Register for business news.

What's still unclear?

Several points are open as of 16 October 2026.

• Single sourcing. The page count, document categories and executive testimony plan trace to one anonymous person cited by Axios. Reuters repeated them without independent confirmation.
• No published FTC document. The confirmation reached the public through a spokesperson speaking to outlets. The demands themselves have not been published.
• When the probe began. CBS News described a probe the FTC launched this summer. TechRepublic said an investigation launched earlier this year. Neither gives a start date.
• Deadlines and scope. No source states response deadlines or which specific incidents the demands name.
• Anthropic's view. No Anthropic statement has been reported.
• Who is next. Axios's source said third-party AI organizations will get demands later. None are named.

The weakest point in any reading of this story is the gap between a document request and a finding. The reported demands show the FTC is gathering evidence. They do not show what it has concluded.