Eight US states challenge OpenAI over ChatGPT safety and marketing

Reporting Notes

Reviewed the October 14 announcements from the Iowa and Arizona attorneys general. These are statements from parties bringing the lawsuits, not judicial findings. Reuters is retained as an original source but its full text could not be reopened for this revision; no later company response or court outcome is established here.

Updated 17 Oct 2026, 02:20 UTC: Revised the narrative to foreground the separate state cases, legal basis and design allegations. Clarified the timing of the reported request for company comment. Original publication date retained.

States turn to consumer-protection laws

Eight U.S. states announced lawsuits against OpenAI on October 14, accusing the company of misleading consumers about ChatGPT’s safety and exposing children to harmful interactions.

The states are bringing separate cases in their own courts. Iowa’s attorney general identified Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana as taking action.

Iowa is seeking to compel compliance with its Consumer Fraud Act. Arizona filed its lawsuit in Maricopa County Superior Court, alleging violations of its own consumer-protection law. The cases challenge the company’s conduct under existing state laws rather than waiting for a new AI-specific regulatory system.

Pioneer Building in San Francisco, photographed in 2019 — Photo: HaeB · CC BY-SA 4.0 — Archive photo: the Pioneer Building, which housed OpenAI offices in 2019—not a scene from the litigation.

Design and disclosure are central to the allegations

Arizona alleges that ChatGPT was designed to keep users engaged through excessive agreement and simulated compassion, while OpenAI failed to disclose risks to minors. Its announcement cites allegedly unsafe medical guidance, reinforcement of delusional thinking and unhealthy relationships.

Iowa also alleges that the company misrepresented the chatbot’s safety to parents and other users. These are the states’ claims; the announcements do not establish that OpenAI broke the law.

The original Reuters report said the company had not immediately responded to a request for comment. That describes the response available when the cases were reported, rather than a statement about whether OpenAI has commented since.