FTC starts inquiry into OpenAI and Anthropic AI risks

The US FTC began an investigation of OpenAI, Anthropic, and other AI makers. It wants to learn what risks their technology may create for consumers.

The review includes the research group METR. The FTC is focusing on AI agents that act on their own. OpenAI and Anthropic hired METR for independent investigations of safety incidents involving these agents.

According to Habr, the FTC plans to send demands to developers in the coming weeks. These are called Civil Investigative Demands, or CID. They legally require developers to provide documents and other information. The FTC also plans to require company leaders to give testimony.

FTC chair Andrew Ferguson had voiced concerns about companies earlier. This was before an incident with OpenAI agents on Hugging Face. He said developers should answer for harm when they give agents instructions for cyber safety tests that lead to hacks. He urged officials to use current laws first, not make new rules for AI.

The FTC has broad power to sue companies for unfair or misleading practices. It used this power against companies that did not take reasonable steps to protect consumer data. On Tuesday, Donald Trump met leaders of leading AI companies. They agreed on voluntary standards. The Independent reported that the meeting took place one day before news of the investigation appeared.