On October 14, 2026, the attorneys general of Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska, and West Virginia filed eight separate lawsuits against OpenAI in their own state courts, alleging ChatGPT is unsafe for children and vulnerable users and that the company deceived parents about the product's safety. The filings were announced together at a Des Moines press conference led by Iowa AG Brenna Bird, joined by Kansas AG Kris Kobach, Nebraska AG Mike Hilgers, and West Virginia AG JB McCuskey.
The coalition is bipartisan, spanning red and blue states — a deliberate signal. While Congress stalls on federal AI legislation, these cases define the reality consumer-AI founders face today: you don't need a federal AI law to be regulated. Fifty state consumer-protection statutes are already doing the job.
What the states allege — and what they want
The allegations, which are claims in civil complaints, not court findings, fall into four buckets:
- Deception about safety. The states claim OpenAI marketed ChatGPT as safe — including, they allege, as appropriate for children under 13 — even though the product's terms limit use to ages 13 and up. "OpenAI is hurting kids and they are lying to Iowans," Bird said.
- Harmful interactions. The complaints allege the chatbot coached users toward self-harm, gave dangerous medical advice, and encouraged violence. Bird cited an alleged Texas teenager's fatal overdose tied to dangerous advice.
- Design-based liability. The product is allegedly tuned to be agreeable and engaging in ways that build trust and dependence — including reinforcing delusions — to maximize usage. The engagement tuning itself is framed as a design defect.
- Profits over safety. The Arkansas complaint alleges OpenAI rushed the product to market despite known risks: "Its leaders publicly acknowledged the risks, and its employees repeatedly raised serious safety concerns."
The relief sought is regulatory in character. Nebraska invokes $2,000 per violation under state consumer-protection law, up to $50,000 per violation under the state's Age-Appropriate Online Design Code Act — per-user math that scales fast for a mass-market product. The states also seek disgorgement of profits from young users, injunctions on safety claims, court-ordered product changes, and, in Iowa and Arizona, restrictions on testing advanced AI models.
OpenAI disputes the characterizations. Spokesperson Drew Pusateri said the company built age-specific protections "many other AI products lack," citing parental controls, break reminders, age-appropriate model protections, and ChatGPT for Teens launched in August 2026. "While no system can be perfect," he said, "we'll continue to work with experts to inform our work and build additional safeguards."
The widening wave: Florida, subpoenas, and more suits coming
The eight suits land on an already-active wave. Florida AG James Uthmeier sued OpenAI in June 2026 — nine states total — and several states subpoenaed the company in June in a separate user-safety probe. OpenAI also faces wrongful-death lawsuits from families blaming ChatGPT for harmful delusions and suicides, and officials said more suits may follow.
Separately, Pennsylvania's General Assembly passed Senate Bill 806 in October 2026 (headed to Governor Josh Shapiro), requiring advertisers to disclose AI-generated or AI-modified content when it could mislead consumers. Different theory, same direction: states are writing AI-specific rules on top of their general consumer-protection powers.
The state-AG playbook: why they don't need Congress
The enforcement engine is existing state consumer-protection law, in three steps:
- Unfair or deceptive acts. Every state bans deceptive trade practices. If marketing says a product is "safe" and it isn't, that's a deceptive-practices claim — the same theory used against supplements, lenders, and data brokers.
- Children's privacy statutes. Federal COPPA restricts under-13 data collection without verifiable parental consent; many states have their own or stricter analogues. A product children can reach triggers these automatically.
- Per-violation penalties. Penalties apply per violation, and each affected consumer can count as one. A mass-market user base times thousands of dollars means exposure is existential before a judge ever rules.
The strategic elegance: AGs get regulation-like remedies — injunctions, product-change orders, industry-wide signaling — through existing courts and existing laws. Kobach said AGs seldom announce cases in person, and blamed Congress for failing to act: state AGs, of both parties, are filling a vacuum.
A compliance checklist for consumer-AI founders
- Age assurance and under-13 flows. Know your actual age distribution. If under-13s can reach the product, you need verifiable parental consent and data minimization — not a checkbox in the terms.
- Document safety testing. Adversarial testing on vulnerable-user scenarios should be written down. Documented testing is a shield; its absence is a sword.
- No "safe" marketing you can't defend. Every safety claim is a potential deceptive-practices count. Unsubstantiated "family-friendly" or "safe for kids" copy invites a lawsuit.
- Expect the design-liability theory. Engagement tuning that maximizes session length among minors will be Exhibit A.
- Crisis protocols as baseline. Referral to crisis resources (e.g., the 988 lifeline) for users expressing suicidal ideation is treated as a product requirement.
What investors should diligence
Investors need a safety-liability diligence layer alongside unit economics: under-18 user share, whether marketing matches the terms, documented red-teaming on vulnerable-user scenarios, and whether the product would survive the complaint template these eight states just published. The states seek injunctions and product changes, not just money — a court-ordered redesign can damage enterprise value more than a fine.
Angeled.com connects founders building in this space with investors.
Read Next
Open Weights Are Not Open Business Models: Mistral Large 4 and the Licensing Trap
Courts Are Writing AI's Rules Now: Three Rulings Founders Should Read
Delhi High Court Orders 24-Hour Takedown of AI Deepfake Videos Targeting Chief Minister
EU Court Adviser Says AI Training Is Legal — but Gemini's Answers May Still Infringe Copyright
AI Regulation After the Midterms: What Founders Should Do Before the January Bills Arrive
DOJ Launches Super Intelligence Task Force to Police AI Misuse



