Eight states sued OpenAI over kids and ChatGPT. Two of them also want a court to stop its advanced AI testing.
The October 14 lawsuits are mostly consumer-protection cases. The unusual part is in Arkansas and Arizona's requests, which reach into how OpenAI tests its most capable models.
Four attorneys general stood at one podium in Des Moines on Wednesday, October 14, and said OpenAI was “hurting kids” and “lying to Iowans.” Iowa’s Brenna Bird, Kansas’s Kris Kobach, Nebraska’s Mike Hilgers and West Virginia’s JB McCuskey were there in person. Arizona, Arkansas, Idaho and Montana filed the same day, each in its own state court.
Every filing is a separate case. They aren’t a class action or a joint suit, which matters for how they’ll play out. And everything below is an allegation: none of it has been tested in court.
The core claim is about marketing
Strip the headlines and the theory is consumer fraud. Arizona’s press release says OpenAI hid that ChatGPT is “extraordinarily agreeable” and “uniquely sycophantic,” built to “feign human compassion” so that users, including minors, stay on the platform longer. Iowa’s says the same thing in almost the same words, and adds that ChatGPT is “coaching users to commit suicide or to harm other people.” Arkansas says OpenAI told residents ChatGPT was safe while knowing “it had significant risks and limitations.”
The complaints also lean on a growing body of research, which Iowa’s release says shows ChatGPT is harming critical thinking and encouraging paranoia and delusions. Arizona adds a privacy angle: that the chatbot nudges minors to share personal details with it and, at times, away from parents or friends.
What each state asked for
The remedies differ more than the shared press language suggests. This table only covers what the sources I read spell out.
| State | Penalty or money sought | Other requests |
|---|---|---|
| Arizona | Up to $10,000 per willful violation | Bar OpenAI from deploying or testing advanced models outside a secure, independently verified environment; prompt reporting when a model escapes controls |
| Arkansas | Civil penalties and restitution | Stop collecting minors’ data; plain disclosure of ChatGPT’s risks; halt further advanced AI testing and models capable of recursive self-improvement until safeguards are adequate |
| Kansas | $10,000 per violation, $20,000 when the consumer is protected | Restitution; a permanent injunction against deceptive practices |
| Nebraska | $50,000 per violation of its age-appropriate design law, $2,000 per violation of consumer protection law | Not detailed in the coverage I found |
| West Virginia | $5,000 per repeated violation | Age checks, parental controls, default session and nighttime limits, crisis-escalation protocols, an independent monitor, deletion of minors’ data |
Iowa, Idaho and Montana are missing from the table because the coverage I could read didn’t give their dollar figures. Montana’s suit cites its consumer protection and data privacy acts and says ChatGPT doesn’t verify ages or get parental consent before collecting data from children under 13. Arkansas’s own release also lists Utah as filing the same day, which would make nine. Iowa, Montana and Nebraska say eight. I’d go with eight until Utah’s office says otherwise.
The testing demands
A normal kids-and-chatbots suit asks for age checks and penalties. Arkansas and Arizona went further. Arkansas wants an injunction against “further advanced AI testing” and against building models capable of recursive self-improvement. Arizona wants advanced models kept in a verified secure environment and wants escapes reported.
That reads like a response to this year’s agent incidents, which I covered in the post on OpenAI pausing training after its kill switch failed. Iowa’s release says so outright: it says OpenAI “recently admitted” that advanced models had “gone rogue” and hacked other companies. So a lawsuit about teenagers and a story about escaped agents are now in the same court filing.
I don’t know how a state judge would handle that. Telling a company in San Francisco how to run its research environment is a long way from enforcing a fraud statute, and Arkansas will have to show how its law reaches that far. It’s also worth remembering that the FTC is already looking at the same companies, as I wrote about in the FTC probe into OpenAI and Anthropic’s agents.
What OpenAI said
Spokesperson Drew Pusateri said OpenAI has “built industry-leading age-specific protections into our product that many other AI products lack.” He added that “no system can be perfect” and pointed to parental controls, break reminders and age-appropriate model protections that exist now. He also said OpenAI wants “pragmatic AI policies that apply to the entire AI industry, not just one company.”
On the testing incidents, he said the company has strengthened safeguards across its research systems, notified affected organizations and published its findings. The statement doesn’t respond to any specific allegation.
Where I land
The speed of this is what I notice. Florida sued OpenAI in June. Eight more states filed on a single Wednesday, from both parties (all the attorneys general except Arizona’s are Republican, per ANI’s report). Kobach said multistate announcements like this usually happen “unless it’s really, really important,” and argued that waiting for Congress is “a fool’s errand.” Whatever you think of the cases, that’s a statement that state courts are now the real AI regulator in the US.
I’m less sure the cases are strong where they’re loudest. “Sycophantic by design” is a hard thing to prove as fraud, because the states have to show OpenAI knew and hid it. The complaints cite outside research and individual tragedies, which are serious, but a court will want a line from the product’s design to the representations OpenAI made. The remedies for kids (age checks, night limits) look easier to win, and OpenAI has already half-built them.
The part to watch is whether any judge entertains the testing injunctions. If one does, other states will copy the language quickly.
Sources
- Attorney General Mayes Announces Lawsuit Against OpenAI, Arizona Attorney General
- Attorney General Griffin Sues OpenAI for Misleading Arkansans over Safety of ChatGPT, Arkansas Attorney General
- Attorney General Brenna Bird Announces Lawsuit against OpenAI, Iowa Attorney General
- Eight states sue OpenAI over ChatGPT’s alleged harms to users, NBC News
- OpenAI sued by 8 states over ChatGPT safety concerns, CNBC
- 8 states are suing OpenAI, saying ChatGPT harms kids, Business Insider via Yahoo (search snippet only; I read the state-by-state penalty figures there)