OpenAI, Anthropic Face Demands Exceeding 50 Pages in Consumer Protection Probe
Key Facts
The U.S. Federal Trade Commission has issued civil investigative demands to OpenAI and Anthropic, each exceeding 50 pages, seeking information on potential violations of consumer protection laws. An FTC spokesperson confirmed the subpoenas to CBS, which said the move was part of an investigation launched over the summer. According to Reuters, citing Axios, the commission is asking the companies to hand over documents concerning the safety of their products and their practices toward users. Civil investigative demands allow the government to gather evidence and investigate ahead of potential litigation, according to Axios, as part of the procedures the commission uses to examine business practices. The agency has the authority to sue companies over unfair or deceptive practices, the legal power relevant to its investigation into potential consumer protection violations.
The documents sought cover security incidents, safeguards, marketing materials and privacy policies, according to Reuters, citing Axios, extending the scope of the demands to how products are presented to consumers and how their information is handled. An OpenAI spokesperson said the company "will cooperate constructively with the FTC" in response to the regulatory action. The security context of the investigation involves incidents during testing of AI models, including models not available for commercial use, according to Reuters' coverage. The news agency cited an earlier interview with FTC Chair Andrew Ferguson in which he proposed holding developers who direct agents during cybersecurity tests that result in breaches liable for the resulting damage. Ferguson said those developers "should be held responsible for any harm they cause" during such tests.
Alongside the federal investigation, states have filed separate lawsuits against OpenAI, including Arkansas, which sought to require the company to clearly and prominently disclose material risks associated with using ChatGPT. According to a statement from the Arkansas attorney general, the relief sought also includes refunds for consumers harmed by the company practices at issue in the lawsuit. The state requested civil penalties against the company, alongside its demands for user disclosures and refunds for those harmed. In California, Attorney General Rob Bonta issued an investigative subpoena to OpenAI as part of the state's existing investigation, according to a statement from its Department of Justice. The department said its formal investigation covers the Hugging Face platform incident, adding California's proceedings to the legal actions the company faces at the state level.
The FTC previously issued orders to 7 companies in a regulatory effort concerning companion chatbots and their effects on children and teenagers. That study targeted bots used for companionship and conversation, according to the commission's announcement, as part of its focus on protecting younger users when interacting with AI applications. The agency explained that the authority used in that effort allows it to conduct broad studies of companies' activities. It said such studies "do not have a specific law enforcement purpose," under a procedural framework separate from the civil investigative demands currently directed at OpenAI and Anthropic. When announcing the companion chatbot study, Ferguson said "protecting children online is a top priority," setting out the commission's priority in that earlier effort.
OpenAI had announced parental controls for ChatGPT on a page dated 29 September 2025, before the current investigation. The company said the tools allow a parent to link their account to a teenager's account, as part of its announced measures to protect users in that age group. The tools also allow settings to be customized to provide an age-appropriate experience, according to the company's description of the options available to parents to manage their children's use of the service. OpenAI acknowledged in the same announcement that its safeguards "are not foolproof," alongside its presentation of account-linking and settings-adjustment capabilities. It added that those safeguards "can be circumvented" when a user deliberately tries to bypass them, a limitation the company cited when introducing its parental protections.
In the next steps of the investigation, the FTC plans to compel officials at AI developers to testify, according to Reuters. The organizations named in that move include OpenAI, Anthropic and research group METR, alongside the commission's document requests to the two companies. At the U.S. administration level, a meeting of the AI team was scheduled for Thursday, according to an FTC spokesperson quoted by Reuters in a report published on 15 October 2026. Jay Clayton chairs the administration's AI team, while Ferguson leads the FTC, which is conducting the consumer protection investigation. Scott Kupor said he expected a charter setting out the team's objectives and responsibilities to be published after the committee's first full meeting, making that document the next expected step in defining the team's remit.