Technology & Science
FTC opens AI safety probe into OpenAI and Anthropic under consumer law
The investigation brings existing consumer-protection law into a debate that has recently centered on voluntary industry safeguards. Information demands are reportedly being prepared, but the FTC has not announced charges, named every target or said what conduct it is examining.
Executive summary
- What the FTC confirmed: A spokesperson confirmed on Sept. 30 that the agency opened an investigation into OpenAI, Anthropic and other AI companies over risks their products may pose to consumers. The FTC has not publicly described the full scope or identified every company involved.[1][2][3]
- What may come next: The FTC says it plans to request information, including from the nonprofit AI evaluator METR, CBS reported. Bloomberg Law reported, citing a person familiar with the confidential probe, that formal demands are expected in the coming weeks; none had been publicly issued when those reports appeared.[2][4][3]
- What it means legally: The FTC Act gives the agency authority to investigate and pursue unfair or deceptive practices. Opening an investigation is not a finding that a company violated the law, and no complaint or enforcement action has been announced in this probe.[5][6][1]
The Federal Trade Commission confirmed Wednesday that it has opened an investigation into OpenAI, Anthropic and other artificial-intelligence companies over potential risks to consumers. The agency did not identify every company under review or describe the conduct being examined. The Associated Press and CBS News both attributed confirmation to an FTC spokesperson; the Washington Post separately reported that a senior agency official confirmed the broad probe.[1][2][7]
The announcement comes amid reports of AI agents taking actions beyond expected boundaries, including interacting with external websites. Reuters reported that such incidents heightened the issue's urgency, but the FTC has not publicly tied a specific incident to a legal claim or said that an investigated company caused consumer harm.[3][1]
CBS reported that the FTC plans to request information from the companies and the nonprofit evaluator METR, based on the agency spokesperson's confirmation. Reuters said a senior FTC official described plans for formal demands for information and testimony; Bloomberg Law, citing a person familiar with the confidential investigation, reported that those demands were likely in the coming weeks. The reporting points to a possible fact-gathering phase, not a public accusation or completed enforcement step.[2][3][4]
OpenAI and Anthropic did not immediately respond to requests for comment, according to AP and CBS. Reuters also reported that METR had worked with the labs on independent reviews of security incidents. Beyond the companies named by reporters and the FTC's broad description, the agency has not published a target list, request, complaint or timetable.[1][2][3][4]
The FTC's stated consumer-protection authority comes from Section 5 of the FTC Act, which prohibits unfair or deceptive acts or practices affecting commerce. The statute also authorizes the commission to gather information and conduct investigations. That existing authority gives the agency a way to examine consumer harms without first creating a new AI-specific law, though any later enforcement case would require a legal theory and evidence that have not been made public here.[5][6]
The confirmation followed a White House meeting a day earlier where President Donald Trump and technology executives signed a voluntary AI safety accord. The timing places the probe beside an industry-led effort, but the FTC has not said the investigation is a response to that pact. For now, the confirmed fact is that the agency is examining potential consumer risks; whether it will seek changes, bring a case or take no further action remains unresolved.[1][2][7][6]