FTC subpoenas OpenAI and Anthropic for safety incident records

The IPO link comes from one place. Cryptobriefing, summarising the Axios report, listed among its takeaways that the heightened regulatory environment could affect OpenAI's strategic decisions, including its IPO plans and funding strategies ^[12]. Its only market datum was a slight rise in the odds of an OpenAI bankruptcy by December 2027 ^[13].

The reports describe a document demand. Two demands of more than 50 pages each come to more than 100 pages of requests ^[11]. The legal footing is the FTC's broad authority to sue over unfair or deceptive practices, which it has used against companies that failed to take reasonable measures to secure consumers' data ^[10]. The probe follows reports that models hacked websites and behaved deceptively or aggressively in testing ^[7]. All of it rests on one person familiar with the matter, via Axios ^[1], and the FTC, OpenAI and Anthropic did not immediately respond to Reuters' requests for comment ^[9].

The inquiry could stop at document production. Or it could produce a complaint under that authority, with Ferguson's view on liability for agent tests as the theory ^[6]. Or the records and testimony could feed the lawsuits and investigations into consumer harms that OpenAI already faces ^[8]. We think the first and third are likelier than the second in the near term, and that none of them reaches a funding round soon. No penalty, deadline or dollar figure has been reported. There is nothing yet to price. The counter-case is cryptobriefing's. Sentiment can move before any filing, and the bankruptcy odds have already ticked up ^[13].

We would change that view if an investor or underwriter delayed or repriced an OpenAI deal and cited the inquiry, or if the FTC filed a complaint against either lab. Until one of those happens, the cost is lawyers' hours at two companies and, if the FTC follows through, executive time spent testifying ^[4].