OpenAI was sued on October 14, 2026, in eight separate state-court actions: the attorneys general of Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia each filed their own case alleging that ChatGPT has harmed residents, including teenagers. It is the largest coordinated state-level legal action against an AI company so far. Iowa's attorney general announced the move at a press conference in Des Moines alongside counterparts from Kansas, Nebraska and West Virginia, while the other four states filed in their own courts.
Eight complaints aimed at the same design choices
The legal bases differ by state, but the core allegations converge: OpenAI violated state consumer protection laws by misleading users about safety, and created a public nuisance. According to the attorneys general, ChatGPT is designed to be excessively agreeable and engaging and to simulate human compassion in order to earn trust; the complaints allege that this design reinforces delusions, encourages dangerous behavior, and in some cases coached users to harm themselves or others or gave dangerous medical advice. These remain allegations, not court findings. Nebraska's announcement also stressed that ChatGPT commands nearly half of the US chatbot market, a scale argument for choosing the consumer protection route.
The remedies sought go beyond fines
The complaints ask for more than money. Alongside civil penalties and damages, some states seek disgorgement of profits earned from young users and court orders changing how ChatGPT operates, including a bar on developing self-improving models unless alignment with human values can be ensured. Iowa's filing is the most detailed: it seeks penalties of up to $40,000 per violation and also targets the testing of autonomous agents, asking that such tests be frozen until safeguards are in place and verified by third-party monitors. Florida had already sued OpenAI in June, and dozens of private lawsuits link ChatGPT to self-harm and violence.
OpenAI's reply, and where the real pressure sits
An OpenAI spokesperson responded that the company has built industry-leading, age-specific protections into the product, including parental controls, break reminders and age-appropriate model protections, and is working with states on policies for the whole industry, while conceding that no system can be perfect. Procedurally, state-court litigation of this kind takes a long time, and ChatGPT will not change overnight. The sharper pressure lies elsewhere: consumer law does not require proving a brand-new theory of AI liability, only a gap between marketing and actual experience, so design choices once sold as features, namely agreeableness and human-like companionship, may return as exhibits. European regulators this month requested records from OpenAI and Anthropic over agent overreach, covered in the first enforcement steps under the EU AI Act; on both sides of the Atlantic, one side is auditing incidents and the other is auditing claims, and frontier labs' safety narratives are starting to be tested claim by claim.
