Policy & law

FTC sends 50-page CIDs to OpenAI and Anthropic over safety incidents

Axios: civil investigative demands dated 14 October. Marketing and privacy files are in scope.

2 min readFrontier Deskftc, OpenAI, Anthropic, policy

Share

The investigation notice was last month's story. The paper that landed this week is the civil investigative demand.

CIDs over 50 pages each, dated 14 October

Axios reported on 15 October 2026 that the FTC on Wednesday (14 October) sent civil investigative demands to Anthropic and OpenAI in a probe focused on potential consumer-protection violations. A person familiar with the matter told Axios each demand is more than 50 pages and seeks documents on security incidents and safeguarding practices, plus marketing materials and privacy-policy documents. Axios says investigative demands will go to third-party AI organizations later, and that the FTC plans to compel top AI executives to testify about their products.

An OpenAI spokesperson told Axios: "As we improve our models and they increasingly help people in their day-to-day lives, safety remains a top priority. We will engage constructively with the FTC." Ars Technica's 15 October report, also grounded in the Axios scoop, says the FTC and the companies did not provide the CID text for review. Ars notes Reuters had reported in September that a probe was coming and that METR could be drawn in; that earlier notice is background, not a second set of CIDs in this article.

Axios also reports that the same Wednesday a group of state attorneys general sued OpenAI under consumer-protection laws over chatbot safety claims involving users including children. That suit is parallel litigation, not the FTC CID itself.

Who must answer the CID

The FTC issued the demands. OpenAI and Anthropic are the named recipients on the Axios account. Future third-party targets and executive witnesses are flagged as planned, not as a published service list. Readers who only saw Frontier's earlier investigation brief should treat this piece as the document-production beat, not a duplicate launch.

From probe notice to document production

Dated civil investigative demands with a reported 50-plus-page scope that explicitly includes marketing and privacy files alongside incident and safeguard records, plus a stated plan for executive testimony. It is not a complaint filed in court. It is not a published finding that either company violated the FTC Act.

The CID PDF is not on the public docket yet

Date: CIDs reported sent 14 October 2026; Axios scoop 15 October. The full question lists, return dates, and confidentiality legends are not in the Axios or Ars accounts. Whether METR or other evaluators have received demands as of this writing is not confirmed in those stories. Penalty exposure if the FTC later sues remains speculative; Chair Andrew Ferguson's prior comments about breach-style notification and agent liability are policy posture, not a fine schedule attached to these CIDs.

Counsel should map incident files to the marketing record

Labs and vendors in the blast radius should inventory safety-incident tickets, eval escape writeups, consumer-facing safety claims, and privacy policies against a single privilege log before any informal FTC call. Buyers who relied on vendor safety marketing should keep those PDFs; they are in the reported CID scope. Do not treat "engage constructively" as a closure date.

The CID PDFs themselves are not linked from these reports.

Share

Cite this piece

Canonical URL

https://www.thefrontier.dev/articles/ftc-cids-openai-anthropic

Attribution

The Frontier, “FTC sends 50-page CIDs to OpenAI and Anthropic over safety incidents”, 16 Oct 2026

TLDR

Axios: civil investigative demands dated 14 October. Marketing and privacy files are in scope.

Plain text

Frontier Desk. “FTC sends 50-page CIDs to OpenAI and Anthropic over safety incidents.” The Frontier. 16 Oct 2026. https://www.thefrontier.dev/articles/ftc-cids-openai-anthropic

BibTeX

@misc{frontier_ftc_cids_openai_anthropic_2026,
  title = {FTC sends 50-page CIDs to OpenAI and Anthropic over safety incidents},
  author = {{Frontier Desk}},
  howpublished = {The Frontier},
  year = {2026},
  month = oct,
  url = {https://www.thefrontier.dev/articles/ftc-cids-openai-anthropic}
}

Full text may be reprinted with canonical link and byline.