States Deploy Social Media Playbook Against OpenAI Before Federal AI Law Passes

Published by James Harris on June 13, 2026June 13, 2026

A coalition of state attorneys general, led by New York, has subpoenaed OpenAI for documents covering ChatGPT’s algorithmic design, data practices, user retention tactics, and how its products interact with children and elderly users. This is not a warning letter. It is a formal legal demand with document production requirements, and it follows Florida becoming the first state to sue OpenAI directly on June 1.

The pattern here is familiar to anyone who watched state regulators move on social media platforms in the early 2020s, or on crypto exchanges before federal frameworks existed. States consistently outrun Congress when federal legislation stalls, and they have. The GUARD Act passed the Senate Judiciary Committee unanimously in late April but has no scheduled floor vote. In that vacuum, 49 states and D.C. have introduced 464 AI-related bills since 2025, according to the National Conference of State Legislatures. What happened with social media is the relevant precedent: companies faced a patchwork of conflicting state obligations, compliance costs ballooned, and some product features were effectively killed by litigation risk before any federal standard was set. OpenAI is now entering that same dynamic, except its core product is a single chatbot used by hundreds of millions of people, which concentrates the regulatory exposure considerably.

Florida’s lawsuit names Sam Altman personally, which is not a standard move and signals that at least one attorney general is trying to pierce the corporate structure entirely.

The subpoena’s scope matters beyond OpenAI. Google, Meta, and other AI providers received letters from the National Association of Attorneys General in December 2025 raising similar concerns, with a January 2026 deadline to document their safeguards. That deadline has passed. If the multi-state investigation produces documents showing OpenAI’s internal safety assessments diverged significantly from its public statements, the resulting disclosures would likely accelerate parallel actions against other AI companies facing the same questions. The children’s safety angle also gives state AGs the most legally durable ground to stand on, since child protection statutes carry strict liability in several jurisdictions and have survived federal preemption challenges that broader privacy claims have not.

OpenAI has said it will engage constructively with the attorneys general. The Florida litigation, which grew from a criminal investigation into chat logs connected to the 2025 Florida State University shooting, is already in court and has no clear near-term resolution.