FTC opens investigation into OpenAI and Anthropic over consumer risks

By Splitfeed EditorialLast updated Oct 2, 2026, 12:30 UTC

On September 30, 2026, the US Federal Trade Commission formally opened investigations into two of the country's most prominent AI developers, OpenAI and Anthropic. The probe centers on possible risks those companies pose to consumers, with at least one framing of the inquiry focusing specifically on whether AI agents, systems capable of taking autonomous actions on behalf of users, can be adequately controlled. A separate detail emerging from the coverage notes a distinction between the two companies' approaches, characterizing Anthropic as applying a hard cap and OpenAI a softer touch, though the precise subject of that contrast was not specified in available reporting.

The timing matters because both OpenAI and Anthropic have been aggressively expanding their AI agent products throughout 2026, moving beyond chatbots toward systems that browse the web, execute code, manage files and interact with third-party services on a user's behalf. An FTC investigation at this stage signals that federal regulators are no longer treating consumer protection questions around AI as hypothetical. Whether the agency has the statutory tools to act on whatever it finds, and what remedies it might seek, remain genuinely open questions. The distinction between a hard cap and a soft touch, however briefly noted, hints that the investigation may scrutinize specific design choices rather than treating the two companies as a single bloc.

None of the three camps have published reactions to this story yet. Pro-AI voices would typically argue that existing safety frameworks and self-regulation within leading labs are sufficient, and that premature regulatory intervention risks slowing beneficial innovation. Anti-AI voices would typically welcome the probe as long overdue, arguing that agentic systems operating with limited human oversight represent a concrete and underexamined consumer harm. The middle ground would typically call for targeted, evidence-based rules that address specific failure modes without treating all AI development as equally risky, and might point to the Anthropic-versus-OpenAI contrast as exactly the kind of granular distinction regulators should be drawing.

The argument will sharpen once the FTC signals what specific practices it is examining and whether it moves toward a formal complaint or closes the investigation. Any enforcement action, consent order or public report from the agency would settle much of the current uncertainty about both the scope of the probe and the FTC's practical authority over AI agent design.