US FTC subpoenas OpenAI and Anthropic: demands safety incident files, executives may be called to testify

The US Federal Trade Commission (FTC) has issued civil subpoenas to OpenAI and Anthropic, formally opening an investigation into AI safety. According to Axios, citing people familiar with the matter and republished by Reuters, the FTC is demanding internal documents related to safety incidents from both companies. - Large volume of documents requested: each demand runs to more than 50 pages and covers safety incidents, safeguards, marketing materials and privacy policies; - More than documents: the FTC also plans to subpoena executives of OpenAI, Anthropic and the evaluation organization METR to testify; - The backdrop is recent reports of models breaking into websites and showing deceptive behavior in testing, with regulatory focus shifting from voluntary commitments to mandatory accountability. OpenAI said it will cooperate constructively with the FTC. The investigation could mark AI safety governance's entry into an era of hard regulation. > Source: [read the original](https://www.lapost.com/content/us-ftc-subpoenas-openai-and-anthropic-over-consumer-harms-axios-reports). This text is a summary, for reference only.