Dozens of state attorneys general served OpenAI a subpoena on June 12, and the timing couldn’t be worse: the company had filed its confidential S-1 just four days earlier, targeting what could be a trillion-dollar IPO. New York’s AG issued the subpoena on behalf of the coalition, seeking documents on advertising, user engagement and retention, handling of consumer and health data, handling of minors and seniors, deep learning models, and model sycophancy.
That last one’s worth noting. “Model sycophancy” — where AI systems tell users what they want to hear instead of what’s accurate — is now a named legal concern in a formal multistate enforcement action. That’s a new compliance category founders building on consumer AI need to price in.
Florida didn’t wait for the coalition. The state filed the first civil lawsuit against OpenAI and CEO Sam Altman personally on June 1, accusing them of endangering children, facilitating criminal conduct, and deceiving parents about ChatGPT’s safety. AG Uthmeier was blunt: “People are getting hurt, parents are getting deceived, and they need to pay for it.” The complaint seeks parental controls and financial penalties; no dollar amount set yet.
Here’s the read-across for operators building on OpenAI’s API: the subpoena scope covers both consumer data handling and AI model behavior, not just UI-level product decisions. If you’re running a consumer app that touches health data or serves minor users, the legal framework being built here will eventually land on you too. COPPA compliance and sycophancy audits are now in the same regulatory sentence.
OpenAI’s response was what you’d expect from a company with a pending S-1: “we take the concerns raised by state attorneys general seriously.” The real disclosure is in the filing.
Nathan Zakhary