FTC Opens Safety Investigation Into OpenAI, Anthropic, Other AI Firms

The move marks a potential shift in federal AI oversight, following a period in which the current administration has favored industry self-regulation over formal enforcement.

The Federal Trade Commission is preparing civil investigative demands targeting OpenAI, Anthropic, and other artificial intelligence companies over potential safety risks posed by their products, an FTC spokesperson confirmed to Axios on Wednesday, September 30, 2026.

Civil investigative demands are formal legal instruments that compel companies to produce documents, data, and testimony. They do not constitute charges or findings of wrongdoing. The FTC has authority under Section 5 of the FTC Act to investigate unfair or deceptive acts or practices, which could encompass claims companies make about the safety and reliability of their AI systems.

FTC Chair Andrew Ferguson is leading the investigation, according to the Axios report. Ferguson, a Republican appointee, has not publicly detailed which specific products or safety claims are under review. The precise scope of the civil investigative demands has not been made public as of the time of publication.

The investigation represents a notable development given the federal government's posture toward the AI sector over the past two years. The current administration has generally declined to impose new binding regulations on AI developers. In a prior White House technology meeting, President Trump indicated opposition to new federal AI safety regulations, a position The Congressional Times reported in Trump Rejects New Federal AI Safety Regulations After White House Tech Meeting.

OpenAI and Anthropic are among the largest AI developers in the United States by reported valuation and model deployment scale. OpenAI, maker of the GPT series of models, and Anthropic, maker of the Claude series, both market their products to consumers and enterprise clients. Neither company has disclosed the revenue figures or internal safety testing records that would be relevant to an FTC inquiry, and those records are not currently part of any public filing.

The FTC has previously examined AI-adjacent issues, including a 2023 report on AI and competition, and has sent orders to major technology companies requesting information on investments in AI firms. However, a direct safety-focused civil investigative demand targeting AI model developers would represent a step beyond those prior information-gathering exercises, if confirmed in final form.

Congress has not passed comprehensive federal AI legislation as of this date. Multiple bills addressing AI transparency, liability, and safety have been introduced in both the Senate and the House during the current and prior Congress, but none has cleared both chambers. The absence of a statutory framework means the FTC is proceeding under its existing consumer protection authority rather than a dedicated AI statute.

The investigation also follows a separate legal action filed by an AI safety nonprofit against OpenAI over autonomous system access, which The Congressional Times covered in AI Safety Group Sues OpenAI Over Autonomous System Access to Hugging Face. That case is a private civil action and is distinct from the FTC inquiry.

Industry groups have consistently argued that premature regulation could disadvantage U.S. AI developers relative to foreign competitors, particularly those based in China. Proponents of oversight argue that absent enforceable standards, companies have limited external incentive to prioritize safety over product deployment speed.

What remains unknown is which specific AI products or safety representations triggered the FTC's investigation, which other companies beyond OpenAI and Anthropic are named, and the timeline for issuing the civil investigative demands. The text of any issued demands, once served, could become part of the public record through subsequent litigation or congressional disclosure requests. A formal FTC complaint or consent order, if one results from this inquiry, would be published on the FTC's public docket at ftc.gov.