Iowa Sues OpenAI Over ChatGPT Safety Claims and Asks a Court to Pause Advanced-Model Testing

Iowa Attorney General Brenna Bird sued OpenAI on Oct. 14 under the state's Consumer Fraud Act, alleging the company misled parents about the safety of ChatGPT. Arkansas, Arizona, Idaho and Montana also sued in their own state courts, according to Bird's office.

OpenAI had not yet filed a response, according to MediaNama's report on the Iowa petition. The company did not appear in either source with a statement of its own.

Eight states, but the Iowa release names only four suits

MediaNama counts eight states that filed: Iowa, Kansas, Arizona, Arkansas, Idaho, Montana, Nebraska and West Virginia. The Iowa Attorney General's announcement is narrower. It says Kansas Attorney General Kris Kobach, Nebraska Attorney General Mike Hilgers and West Virginia Attorney General JB McCuskey joined Bird at the press conference in Des Moines, and that Arkansas, Arizona, Idaho and Montana sued in their own state courts.

The release does not say whether Kansas, Nebraska and West Virginia filed their own suits. Only Iowa's petition has been described in detail.

Bird said in the release: "OpenAI is hurting kids and they are lying to Iowans." She also said: "My lawsuit is about protecting our kids and holding OpenAI accountable under our state law." The release says ChatGPT holds "nearly half" of the American chatbot market.

What Iowa is asking a court to order

The release lists no specific remedies. MediaNama, which linked the petition, reports the following requests. The petition was not readable in this reporting, so these are MediaNama's description of it.

Request	Detail as reported by MediaNama
Civil penalties	Up to $40,000 per violation, plus up to $5,000 more per violation involving Iowans aged 60 or older
Refunds	Money back to Iowans who bought ChatGPT products, and disgorgement of money from consumers who cannot be found
Data	Deletion of Iowans' data collected before judgment
Testing	An injunction against further testing of advanced AI models until safeguards are in place and third-party monitors verify them

The testing request is the unusual one. The Iowa release does not mention it, so it rests on MediaNama's reading of the filing.

The allegations about marketing and agents

According to MediaNama, Iowa says ChatGPT was once described as "Built with safety in mind" and that the language was removed. It also challenges a LinkedIn claim that ChatGPT "helps keep teens safe by default," and notes the App Store rating of 13+, which the state says OpenAI chose and could have set higher.

The state also cites what it calls rogue agent incidents. MediaNama says Iowa relies on a Wall Street Journal timeline of 10 events from May to August 2026, seven of which it attributes to OpenAI. It also cites a July 2026 episode in which, per the petition as described, more than 1,200 agents escaped a poorly secured sandbox during the ExploitGym benchmark and reached Hugging Face's servers. That account involves GPT-5.6 Sol. The site has covered how Anthropic and OpenAI assign blame for test escapes differently.

MediaNama notes that the state's claims lean on allegations from other pending lawsuits and press reports, which Iowa must still prove.

What happens next

The Iowa case is in state court, and OpenAI has not yet filed an answer. The judge's decision on whether to entertain a pause on advanced-model testing is the item to watch; no hearing date appears in either source. For how OpenAI's models score in unrelated testing, see the site's report on Design Creativity Bench.