WASHINGTON — The Federal Trade Commission has opened an industry-wide inquiry into potential consumer risks from advanced artificial-intelligence systems, with Anthropic, OpenAI and the nonprofit research group METR among the organizations expected to receive information requests.
A senior FTC official confirmed the inquiry to Reuters on September 30, 2026. CBS News separately reported that an agency spokesperson confirmed the investigation and said officials are examining whether company conduct could violate the FTC Act, the federal law that prohibits unfair or deceptive acts or practices and supports fair competition.
The inquiry is an evidence-gathering step, not a finding that any company broke the law. No charges or enforcement order had been announced as of the reports reviewed for this article.
What the FTC plans to examine
The agency plans to issue formal demands for information and may compel testimony from executives at leading AI developers, Reuters reported. METR, which conducts independent evaluations and safety research on advanced AI systems, is also expected to receive a request.
The focus is on possible risks to consumers from increasingly autonomous “agentic” systems — AI tools that can take sequences of actions, use software and interact with online services with less step-by-step human direction. Those capabilities may be useful, but they also raise questions about cybersecurity, authorization and responsibility when a system acts outside intended boundaries.
Reuters reported that the inquiry gained urgency after incidents involving AI agents and cybersecurity testing. The FTC official said the agency is considering how existing consumer-protection authority applies when an AI developer’s system causes harm or when safety claims do not match a product’s actual behavior.
Why the FTC Act matters
The FTC does not need a new AI-specific statute to investigate every potential problem. Section 5 of the FTC Act gives the commission authority over unfair or deceptive practices, and the agency has previously used existing authority in cases involving data security, privacy and product claims.
For AI companies, that can make several questions relevant: Were safety capabilities described accurately? Were foreseeable risks addressed with reasonable safeguards? Were consumers or business customers misled about how autonomous a system was? Did the company maintain controls appropriate to the access granted to its tools?
An inquiry can lead to no action, voluntary changes, negotiated settlements or a later enforcement case. The current reporting does not establish which outcome, if any, will follow.
Companies have not publicly responded
Anthropic, OpenAI and METR did not immediately respond to Reuters requests for comment. CBS News also reported that Anthropic and OpenAI had not responded to its requests by publication time.
The investigation follows a White House meeting at which technology leaders signed a voluntary AI safety accord calling for layered controls, audits and recurring coordination. Voluntary commitments and federal enforcement serve different functions: companies may choose to adopt the former, while the FTC can use statutory authority to demand information and pursue violations.
What to watch next
The next concrete step is the issuance of civil investigative demands or other formal requests. Those documents should clarify the products, time periods and practices under review. Testimony or document production could also show whether the inquiry remains industry-wide or develops into company-specific matters.
Consumers should not interpret the announcement as proof that a particular AI product is unsafe. The verified development is that the FTC is gathering information about potential risks and possible legal violations. Any conclusion will require evidence developed through the investigation.
Sources
Featured image: AI processors, used illustratively. Photo by Igor Omilaev via Unsplash.
[…] regulators examine how AI companies manage consumer and public risks. AskNovus recently covered the FTC inquiry involving OpenAI, Anthropic and other AI labs. For a separate product-development story, see the restricted rollout of Google’s Gemini 4 […]