The Federal Trade Commission has served civil subpoenas on OpenAI and Anthropic seeking internal records on a string of artificial intelligence safety incidents, escalating federal scrutiny of the industry’s most prominent developers after months of reports that experimental AI agents behaved in deceptive or aggressive ways.
The civil investigative demands, first reported by Axios, were sent Wednesday and run to more than 50 pages each, according to a person familiar with the matter. They request documents on security incidents and safeguards, along with marketing materials and privacy policies.
From voluntary oversight to compulsory process
Civil investigative demands are a formal, legally enforceable evidence-gathering tool the FTC typically uses before deciding whether to bring a case. The agency’s consumer-protection authority allows it to pursue companies over practices deemed unfair or deceptive, including misleading claims to customers about how a product works or how data is protected.
The agency said last month that it was opening an industry-wide inquiry and planned to compel testimony from senior executives at leading labs as well as METR, an independent research group the companies have used to investigate incidents involving autonomous AI agents. Similar demands are expected to go to third-party AI organizations later, the person familiar said.
FTC Chairman Andrew Ferguson has argued that developers who direct AI agents in cybersecurity testing that results in real-world hacks should be held liable for the resulting harm.
States open a second front
The federal move coincides with a wave of state litigation. On Wednesday, attorneys general in Arkansas, Arizona, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia filed separate consumer-protection lawsuits against OpenAI, alleging the company misled families about ChatGPT’s safety and exposed minors to harmful content and advice. The suits seek injunctions, civil penalties and restitution.
The pressure builds on several administration actions aimed at AI developers, including a new incident-notification mandate from the White House’s Super Intelligence Force, which Ferguson co-chairs and which is scheduled to meet Thursday.
OpenAI, Anthropic and the FTC did not immediately respond to requests for comment on the subpoenas, according to Reuters.
Sources & Credits
- Axios: Scoop: FTC subpoenas OpenAI, Anthropic over consumer harms
- Reuters via The Globe and Mail: U.S. FTC subpoenas OpenAI, Anthropic over safety incidents, report says
- Reuters: FTC opens probe into AI giants including Anthropic and OpenAI
- NBC News: Eight states sue OpenAI over ChatGPT’s alleged harms to users
Image: G. Edward Johnson — FTC Building sign, Washington DC via Wikimedia Commons, CC BY 4.0.
Join the conversation