OpenAI got hit last Friday with a sweeping subpoena from a New York-led coalition of state attorneys general that is digging into how its chatbots attract and retain users, how the company treats consumer and health data, and what protections exist for vulnerable groups, including children and seniors. The order, described by reporters as wide-ranging, demands slides, metrics, internal notes, and memos that could reveal how product design, marketing, and safety decisions have been made inside the company. For a firm that sits at the center of everyday digital tools and high-stakes court fights, it is a clear escalation in state-level scrutiny.
As reported by Reuters, the subpoena, sent by New York’s attorney general, seeks material on advertising, user engagement and retention, handling of consumer and health data, activities involving minors and seniors, deep-learning models and internal company policies. Coverage that reviewed the document also highlighted language aimed at probing whether model incentives steered outputs toward flattering or dangerous responses, a concern some outlets have summed up as model sycophancy.
Timing and backdrop
The probe lands just days after OpenAI said it had filed a confidential draft S-1 with the Securities and Exchange Commission, a move the company described as giving it the option to go public while keeping the timeline open, according to OpenAI. It also follows a state lawsuit filed June 1 by Florida’s attorney general that names OpenAI and CEO Sam Altman and accuses the company of marketing a product that has caused real-world harms, per reporting by the Associated Press. Taken together, the mounting legal pressure and the confidential IPO paperwork set up a tense timeline that regulators, plaintiffs and investors are almost certain to track closely.
What investigators are asking for
According to reporters who have reviewed the subpoena, investigators want product road maps, testing and safety-team notes, retention and ad-targeting metrics and communications about model changes, material that could show what OpenAI knew about risky outputs and when. The requests are broad enough to sweep up engineering presentations, marketing decks and internal policy memos, per Engadget. That mix of technical details and commercial strategy is the kind of internal paper trail state enforcers often rely on to piece together how a company actually operates.
Why states use multistate coalitions
Coordinated actions by state attorneys general have long been a go-to tactic for tackling nationwide consumer-protection issues, from the 1998 tobacco Master Settlement Agreement to later waves of litigation over opioids and big tech, because they pool resources and can demand sweeping remedies, according to the National Association of Attorneys General. NAAG notes that multistate efforts can yield both large financial settlements and injunctive changes in industry practices, a combination that helps explain why states often move together when they see emerging public health risks.
What comes next
OpenAI has said it “takes the concerns raised by state attorneys general seriously” and intends to engage constructively with investigators, according to statements quoted by news outlets. Reuters and others report that the next phase is likely to involve extensive document production and review; multistate probes commonly stretch on for months or years and can end in anything from negotiated agreements to full-blown litigation. For OpenAI, any findings that point to gaps in safety or disclosure could complicate its road to the public markets and reshape how AI tools are pitched to minors and seniors.
