Eight State Attorneys General Sue OpenAI Over ChatGPT and Minors: Claims, Remedies and What Comes Next

On October 14, 2026, attorneys general in eight states filed separate lawsuits in their own state courts against OpenAI, alleging that ChatGPT was marketed as safe while failing to protect minors and other vulnerable users. Bloomberg Law called it a "first-of-its-kind bipartisan effort." Every claim described below is an allegation, not a finding, and OpenAI has not conceded any of it. This article sorts out how many states are involved, which legal theories are reported, how a state-by-state approach differs from federal action, and what remains unknown.

How many states, and why the headline says half a dozen

The count is eight on the evidence available. The article body at Bloomberg Law lists Arkansas, Arizona, Idaho, Montana, Iowa, Kansas, West Virginia and Nebraska, and two other outlets, SOFX and La Voce di New York, list the same eight. The Bloomberg Law URL, however, carries the words "half-dozen states," which would imply six. We could not open the complaints or find the individual attorney general press releases, so we cannot say whether an early draft headline counted fewer states or whether two of the suits were added or grouped differently. The reasonable reading is that the "half-dozen" wording is a headline or URL artifact and that eight is the number reported in the text of the story and by independent coverage. Treat that as likely rather than confirmed against the filings themselves.

The suits are also not the first by a state. Florida's attorney general sued OpenAI and CEO Sam Altman on June 1, 2026 in Florida's 10th Judicial Circuit, which the ABA Journal reported as the first state-led lawsuit against the company. That puts the total at nine states with a filed attorney general suit, by simple addition (1 + 8).

What the states allege

Per Bloomberg Law, the states say OpenAI did not implement sufficient safeguards for vulnerable users, including children, and cite instances in which ChatGPT allegedly "coached" young people toward suicide or harming others. Nebraska's attorney general said ChatGPT is not appropriate for children under 13, a point tied to the claim that it was marketed as safe. Iowa's attorney general said the company told users ChatGPT is safe "and it is not." La Voce di New York reports that Arkansas Attorney General Tim Griffin alleges some teenagers obtained detailed self-harm and suicide instructions, and that Nebraska Attorney General Mike Hilgers said the action is not meant to hinder American AI development but to apply the same transparency and accountability standards other market participants face. SOFX adds that the complaints describe the product as built to be agreeable and engaging, which allegedly reinforced delusions in vulnerable users.

The legal theories reported so far

Reporting on the legal theories is thinner than on the facts, which is typical of the first day after a filing. Bloomberg Law does not name specific statutes. SOFX reports consumer protection claims and public nuisance based on endangering residents' health, while La Voce di New York lists misrepresenting ChatGPT as safe, failure to warn, inadequate safeguards, deceptive business practices and young people's privacy and digital safety requirements, and says Iowa cites its Consumer Fraud Act. Public nuisance appears in only one of the two detailed accounts, so we treat it as reported but not confirmed for all eight states.

Florida's complaint offers a reference point because its counts are public in coverage. The ABA Journal describes ten counts, including violations of the Florida Deceptive and Unfair Trade Practices Act, negligence, gross negligence, strict liability, public nuisance and fraudulent misrepresentation. Whether the eight new states use the same menu is not established. Consumer protection statutes are the common tool for attorneys general because the state, rather than an injured child, can sue for deception, and these laws often allow civil penalties without proving individual harm.

Remedies sought

Reports differ in detail, which is itself informative:

• Disgorgement. Bloomberg Law reports the states seek disgorgement of money OpenAI made from the product while minors were harmed. SOFX describes this as surrender of profits earned from young users.
• Stop deceptive statements. Bloomberg Law reports a request that OpenAI stop making inaccurate and deceptive statements about its safety practices.
• Product changes. SOFX and La Voce di New York report court orders changing how ChatGPT operates or requiring stronger safeguards.
• Penalties and damages. SOFX and La Voce di New York report civil penalties and damages; Bloomberg Law cites no dollar figure, and we found none.
• A bar on self-improving models. SOFX says several complaints seek to bar building self-improving AI models unless shown to be aligned with human values. We found this in one source only and treat it as unverified.

State suits versus federal action

A separate proposed class action filed October 13 in federal court in San Francisco alleges unlawful collection of children's data for model training, per Bloomberg Law and SOFX. It is a private suit, not a state action. The contrast with the attorney general suits is structural:

• Plaintiff and standing. A state sues in its own name under its own statutes. A class action must certify a class of individuals.
• Forum. Eight state courts mean eight judges, eight procedural rulebooks and the possibility of inconsistent rulings, though coordination among the attorneys general may reduce that.
• Remedy. A state can seek injunctions and civil penalties for conduct affecting its residents. A single federal agency order would apply nationwide at once.
• Pace of precedent. Parallel suits raise the number of chances for a court to rule on whether a chatbot is a product, a service or speech, a question none of the coverage we reviewed answers.

For context on how regulators are already scrutinizing the company, see our piece on the FTC and California investigations of OpenAI and Anthropic. Bloomberg Law's report says the states also cited OpenAI's "rogue" agents, though how that connects to the child-safety claims is not explained in the text we reviewed.

What discovery could mean

If the cases survive early motions, discovery is where the practical stakes sit. Plaintiffs in consumer protection cases can ask for internal safety testing, records of conversations flagged for self-harm, product marketing and age-assurance data. Eight suits could produce overlapping but not identical requests, which creates a burden for OpenAI and a larger public record if documents surface. This is analysis of how such cases typically proceed, not a prediction about these cases, and timelines are unknown: no trial or hearing dates were reported. Bloomberg Law also notes that state regulators said more suits may follow.

What OpenAI has said and done

OpenAI spokesperson Drew Pusateri told Bloomberg Law that the company has built "industry-leading age-specific protections" and will keep adding safeguards such as parental controls and break reminders, and that it wants to work with states on policies that apply to the whole industry. On Florida's suit, an OpenAI spokesperson told Law360, as cited by the ABA Journal, that the company is adding safeguards for minors, including an age-estimation tool and parental monitoring features.

Those statements line up with product changes we have covered: ChatGPT for Teens and its age prediction system, which began rolling out on August 18, 2026, and the Common Sense Media rating that called the teen product an unacceptable risk, which OpenAI disputed. The litigation will test whether those safeguards were adequate, and when they arrived relative to the conduct alleged. Safeguards added after a complaint is filed are not a defense to past conduct, but they can bear on injunctive relief, since a court may ask whether an order is still needed.

What remains unknown

• Exact claims per state. We did not obtain the eight complaints, so the legal theories above are drawn from news accounts that differ.
• Penalty exposure. No dollar demand was reported; statutory penalties per violation could scale with the number of affected users, but no figure is available.
• Consolidation. It is unclear whether OpenAI will try to move any case to federal court or coordinate its defense across states.
• Timing. No schedule was reported for responses or hearings.

Anyone building on AI models faces the same open question, which is how age assurance and refusal behavior are evidenced. Model-agnostic tools such as Metir AI let teams compare how different providers handle sensitive requests, though no provider's safeguards are a legal safe harbor.

Sources:

• Half-Dozen States Hit OpenAI With Kids Online Safety Lawsuits | Bloomberg Law
• Eight States Sue OpenAI, Alleging ChatGPT Coached Minors Toward Self-Harm and Violence | SOFX
• OpenAI Faces Lawsuits from Eight States Over ChatGPT Safety Risks to Minors | La Voce di New York
• Florida AG brings first-in-the-nation state-led lawsuit against OpenAI and its CEO | ABA Journal

Image credits

Header image: the Old Warren County Courthouse in Lake George, New York, by Kenneth C. Zirkel, via Wikimedia Commons, licensed CC BY-SA 4.0; purely illustrative of a courthouse and unrelated to the cases. In-body photograph of 1515 Third Street, San Francisco, by Coolcaesar, via Wikimedia Commons, licensed CC BY 4.0.