OpenAI faces Iowa bid to stop advanced AI testing until monitors sign off

Iowa Attorney General Brenna Bird asked a state court Wednesday to block OpenAI from continuing to test advanced AI models until third-party monitors can verify the tests are safe.

The request is part of a lawsuit Iowa filed under the state’s Consumer Fraud Act against the maker of ChatGPT.

Bird seeks up to $40,000 per Consumer Fraud Act violation

The petition filed in Iowa District Court for Polk County names eight OpenAI entities as defendants. It asks the court to freeze the testing “until adequate safeguards are implemented to ensure that the tests do not pose a danger to the public and until third-party monitors have verified the same.”

Iowa also wants all Iowans’ pre-judgment data deleted and destroyed by OpenAI. The penalty sought is up to $40,000 for each violation of the Consumer Fraud Act and an additional penalty of up to $5,000 for each violation against an older person.

The petition would preclude OpenAI from collecting and selling Iowans’ personal data without age verification and express, informed consent. It also seeks refunds for customers in Iowa, disgorgement of gross receipts and any additional safety measures the court deems necessary.

Kris Kobach of Kansas, Mike Hilgers of Nebraska and JB McCuskey of West Virginia, all state attorneys general, joined Bird at the press conference in Des Moines. Arkansas, Arizona, Idaho and Montana have all sued OpenAI in their own state courts. According to Cryptopolitan, Florida was the first state to sue OpenAI, back in June.

“OpenAI is hurting kids and they are lying to Iowans,” Bird said. ChatGPT, which has almost half of the US chatbot market, was made to be too agreeable and to simulate human compassion, according to her office.

Drew Pusateri, a spokesperson for the company, said OpenAI has “built industry-leading age-specific protections” into ChatGPT. The tools have parental controls, break reminders and age-appropriate model protections.

Petition pins seven of 10 rogue AI events on ChatGPT’s maker

Another count looks at how OpenAI has been testing AI agents that can work on their own. The petition says that OpenAI was responsible for seven of the 10 “rogue AI” events that happened between May and August 2026. This includes all five breaches in internal testing.

The most well-known one happened in July. During a cyber benchmark called ExploitGym, agents got out of their sandbox and hacked Hugging Face. Cryptopolitan wrote about this event when California subpoenaed OpenAI.

According to the petition, more than 1,200 agents worked as a “collective” and carried out 17,000 actions on Hugging Face servers. GPT-5.6 Sol was one of the models used.

The filing says that OpenAI didn’t know about the activity until after the FBI was notified.

Bird’s office said OpenAI admitted its advanced models had “gone rogue.” The petition also quotes OpenAI chief scientist Jakub Pachocki, who wrote that no lab “has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer.”

If you're reading this, you’re already ahead. Stay there with our newsletter.