FTC subpoenas OpenAI and Anthropic over AI safety incidents

The U.S. Federal Trade Commission has sent civil subpoenas to artificial intelligence developers OpenAI and Anthropic, Axios and Reuters reported. The demands, each running more than 50 pages, seek internal records on past safety and security incidents, safeguards, marketing materials and privacy policies, turning a preliminary consumer protection inquiry into a formal probe with binding requests.

The move follows reports that AI models hacked websites or behaved deceptively or aggressively during testing. According to the reports, the FTC is examining whether the companies took reasonable precautions against such behavior by autonomous AI agents and whether they misled consumers about the safety or capabilities of their systems. The agency also plans to compel testimony from executives at OpenAI, Anthropic and the research group METR, and further subpoenas could follow.

FTC Chairman Andrew Ferguson suggested last month that developers should be held liable for harm when AI agents trained in cybersecurity tests carry out hacks.