The lawsuits do not ban ChatGPT. Instead, eight states are asking judges to order changes to how it works. On October 14, 2026, attorneys general in Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho, and Montana filed separate lawsuits against OpenAI, as Reuters and The Verge report.
The states accuse OpenAI of overstating ChatGPT’s safety, designing it to keep young users engaged in harmful ways, and failing to protect minors. These are allegations, not findings. Their demands run from age verification and nighttime limits to pausing some testing of advanced AI models. One request, from Iowa, would require OpenAI to destroy data it holds on Iowans, adults included.
States are using consumer-protection and youth-privacy laws already on their books to challenge an AI chatbot’s marketing, data practices, and model testing at the same time. For families, the useful question is how today’s ChatGPT safeguards compare with what the states want imposed.
What the states say OpenAI did wrong
OpenAI launched ChatGPT, its conversational AI, on November 30, 2022. The company says it can explain concepts, draft messages and help debug code. According to OpenAI’s help center, teens can use it on eligible Free and paid personal accounts.
Each attorney general built a separate case rather than joining one lawsuit, as the Iowa attorney general’s announcement makes clear. That means the legal claims vary. Iowa Attorney General Brenna Bird alleges consumer fraud in how ChatGPT is marketed and designed. Arizona Attorney General Kris Mayes alleges consumer fraud through safety claims, manipulative engagement and undisclosed risks to minors. Arkansas Attorney General Tim Griffin adds negligence, unlawful handling of young people’s data and public nuisance, a claim that conduct harms the public at large. West Virginia Attorney General JB McCuskey pairs public nuisance with deceptive practices.
Other states lean on privacy and design laws. Montana Attorney General Austin Knudsen alleges consumer-protection and privacy violations, including improper collection of minors’ data. Nebraska Attorney General Mike Hilgers cites the state’s age-appropriate design law, which covers how online services are built for young users. Idaho Attorney General Raúl Labrador points to misleading safety claims, harmful outputs and risks from advanced models.
Several complaints go beyond the chatbot people use every day. Iowa alleges that OpenAI’s testing of advanced models is unsafe, and Arizona, Arkansas, Idaho and Montana seek verified testing conditions or a pause. That concerns how OpenAI develops new models, which is a separate issue from ordinary ChatGPT use.
Kansas Attorney General Kris Kobach’s 72-page petition alleges six violations of the Kansas Consumer Protection Act, according to KZRG. It alleges that OpenAI marketed ChatGPT as safe for teenagers despite evidence that it could encourage suicide, reinforce dangerous delusions and give harmful medical advice, and that it collected data from children under 13 without adequate age verification. Kansas asks a judge to permanently bar the alleged practices and seeks restitution and civil penalties of up to $10,000 per violation, or $20,000 where protected consumers are involved. InfoSeely could not access the petition directly.
What each state wants a judge to order
Requests overlap but are far from identical.
The table lists selected requests, not awards or automatic payouts. Civil penalties are payments sought by the state. Restitution means returning money to consumers; damages means compensation for harm.
| State | Selected requests | Court |
|---|---|---|
| Arizona | Safety measures, an independently verified testing environment, incident reporting, restitution, up to $10,000 per willful violation | Maricopa County Superior Court |
| Arkansas | Disclosures, limits on unlawful youth-data collection and use, a testing halt until independently verified safeguards exist, limits on self-improving models, restitution, penalties and damages | Faulkner County Circuit Court |
| Idaho | Age verification, limits on harmful and addictive features, verified testing conditions, incident disclosure, restitution, $5,000 per violation | Fourth Judicial District Court, Ada County |
| Iowa | Destruction of Iowans’ data collected before judgment, age verification, notice and consent for some data practices, a testing halt until safeguards are verified, refunds, up to $40,000 per fraud violation | Iowa District Court for Polk County |
| Kansas | Permanent ban on the alleged practices, restitution, civil penalties up to $10,000 per violation ($20,000 for protected consumers), per KZRG | Not yet confirmed |
| Montana | Deletion of unlawfully acquired personal data, age verification and privacy compliance before data processing, data security, verified testing, $10,000 consumer-protection and $7,500 privacy penalties per violation | First Judicial District Court, Lewis and Clark County |
| Nebraska | Orders to comply with the law, $2,000 consumer and $50,000 design-code penalties per violation, restitution and damages | Buffalo County District Court |
| West Virginia | Deletion of unlawfully collected minors’ data, age assurance, default session and nighttime limits, parental controls, crisis protocols, independent monitoring, a youth-support plan, financial relief | Mason County Circuit Court |
Iowa’s requested remedies ask the court to make OpenAI “delete and destroy all Iowans’ data obtained prior to the entry of this Court’s judgment.” If granted as written, that would cover adults as well as children.
Arkansas’s filed complaint asks the court to stop OpenAI creating models capable of recursive self-improvement until it can ensure alignment with human values. This means AI that can repeatedly improve its own capabilities.
West Virginia’s requested safeguards include age assurance, meaning checks on a user’s likely age, plus session and nighttime limits switched on by default.
OpenAI spokesperson Drew Pusateri told The Verge in an emailed statement that the company has “built industry-leading age-specific protections” into ChatGPT. He added: “While no system can be perfect, we’ll continue to work with experts to inform our work and build additional safeguards, such as the parental controls, break reminders and age-appropriate model protections that exist today.”
Can parents read their teen’s ChatGPT conversations?
No. OpenAI’s help page says: “Linking accounts does not give a parent or guardian access to a teen’s conversations, chat history, or real-time activity.”
Setup requires both sides. Either a parent or a teen can send an invitation by email or phone number, and the teen must accept it. OpenAI announced these controls on September 29, 2025, more than a year before the lawsuits. Once the accounts are linked, a parent can set quiet hours, when the teen can’t use ChatGPT. Parents can also adjust settings for model training, memory, voice and image generation.
Unlinking takes only one person. Either the parent or the teen can end the link, and OpenAI notifies the parent if the teen does. Safety notifications cover a limited set of situations, so they are not real-time monitoring and won’t flag every concern.
Put next to West Virginia’s demands, the gap is specific. West Virginia seeks default session and nighttime limits. OpenAI says quiet hours are off by default and must be scheduled for a linked teen. Parental controls already exist, but they depend on the teen agreeing to them.
Parents comparing ChatGPT with other apps can apply similar questions from our guide to AI chatbot risks for families, which looks at a different chatbot.
How to tell claims from court orders
The complaints contain allegations and requests for relief. Court decisions and product changes are separate developments.
- Allegation. What a state says OpenAI did wrong.
- Requested remedy. What the state asks a judge to order, as in the table above.
- Judicial decision. A ruling on liability or an order to change something.
- Implemented change. What OpenAI actually changes in ChatGPT, whether ordered or voluntary. That needs its own evidence.
Businesses should distinguish the states’ requested remedies from their existing obligations and any verified court orders. The cases still raise questions worth putting to any AI provider: what user data is kept, for how long, and how a user’s age is checked.
Watch for orders that specify who is covered, what OpenAI must change and when. Those details will decide whether a ruling affects adult data, teen accounts or model testing.
