US REGULATOR SEEKS OPENAI AND ANTHROPIC RECORDS IN SAFETY INQUIRY

The US Federal Trade Commission has issued civil subpoenas to OpenAI and Anthropic seeking records on artificial intelligence safety incidents, according to a person familiar with the matter. The person said the agency sent the demands on Wednesday as part of an inquiry into potential consumer harms. The requests seek details of incidents during testing, including accounts of AI models hacking websites and breaching government systems. The inquiry also concerns claims that models acted deceptively or aggressively during tests.

According to the person familiar with the matter, the demands contain more than 50 pages of questions covering security incidents and safeguards. They also seek privacy policies and marketing materials relevant to whether the companies misled consumers. The FTC has authority to bring legal action over deceptive or unfair practices and has previously used those powers against companies over failures to protect consumer data. Andrew Ferguson said developers whose instructions to AI agents during cybersecurity tests led to hacks should face liability for resulting harm.

The FTC intends to compel executives at OpenAI, Anthropic and the research group METR to testify. The person familiar with the matter said the agency also plans to subpoena third parties targeted by the companies’ AI agents. An OpenAI spokesperson said the company “will engage constructively with the FTC and continue working with regulators and policymakers on policies that raise the safety bar.” The FTC and Anthropic did not immediately respond to requests for comment on the demands.