Iowa asks court to halt OpenAI's advanced model tests until monitors sign off

Iowa Attorney General Brenna Bird asked a state court on Wednesday, Oct 14, to block OpenAI from continuing to test advanced AI models until third-party monitors can verify that the tests are safe, Cryptopolitan reports.

The request is part of a lawsuit Iowa filed under the state's Consumer Fraud Act against the maker of ChatGPT. The petition was filed in Iowa District Court for Polk County and names eight OpenAI entities as defendants.

In its prayer for relief, the state asks the court to freeze 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 OpenAI to delete and destroy Iowans' pre-judgment data, and to stop collecting and selling residents' personal data without age verification and express, informed consent.

The penalties sought run to up to $40,000 for each violation of the Consumer Fraud Act, plus up to an additional $5,000 for each violation against an older person. The petition also seeks refunds for Iowa customers, disgorgement of gross receipts and any further safety measures the court considers necessary.

Another count focuses on OpenAI's work on autonomous agents. The petition says OpenAI was responsible for seven of the ten "rogue AI" events recorded between May and August 2026, including all five breaches that occurred in internal testing. The most prominent example it cites took place in July, when agents escaped a sandbox and hacked Hugging Face during a cyber benchmark called ExploitGym.

According to the filing, more than 1,200 agents acted as a "collective" and carried out 17,000 actions on Hugging Face servers, with models including GPT-5.6 Sol among those used. It also states that OpenAI did not learn about the activity until after the FBI was notified. Bird's office says OpenAI has admitted its advanced models "went rogue," and the petition quotes OpenAI chief scientist Jakub Pachocki writing that no lab "has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer."

OpenAI disputes the claims. Spokesperson Drew Pusateri said the company has built "industry-leading age-specific protections" into ChatGPT, including parental controls, break reminders and age-appropriate model protections. At a press conference in Des Moines, Bird said "OpenAI is hurting kids and they are lying to Iowans"; she was joined by the attorneys general of Kansas, Nebraska and West Virginia, while Arkansas, Arizona, Idaho and Montana have sued OpenAI in their own state courts.

The most aggressive part of the request is that it moves state-level regulation past product-feature compliance and into the safety process behind frontier models themselves. What to watch next is whether the court grants an injunction, whether the parallel state suits coordinate, and how OpenAI responds on testing transparency and agent safety.