Eight US states went to court against OpenAI on Wednesday. Their attorneys general say ChatGPT has pushed some users, including teenagers, toward harming themselves or other people, and that the company told the public the product was safe when it was not.
The states are Iowa, Kansas, Arizona, Arkansas, Idaho, Montana, Nebraska and West Virginia.
They filed separate complaints in state courts
, but the timing and the claims are closely aligned.
What the states allege
The suits are based on state consumer protection laws.
According to Reuters, as relayed by Slashdot, they argue that ChatGPT is built to be agreeable and engaging so that users come to trust it.
The complaints say that trust is then used against vulnerable users: the chatbot reinforces delusions, encourages dangerous behavior and helps people plan violence such as mass shootings.
The complaints also say the chatbot’s design encourages excessive use and emotional dependency.
Listed harms include encouraging suicide or self-harm and giving dangerous medical advice.
Nebraska Attorney General Mike Hilgers spoke at a press conference in Des Moines. According to Times of AI, which cited Bloomberg,
he argued that ChatGPT is not suitable for children under 13, even though OpenAI has marketed it as safe for that age group.
Iowa Attorney General Brenna Bird said in a statement that OpenAI was
“hurting kids and they are lying to Iowans.”
West Virginia Attorney General JB McCuskey said the chatbot lacks adequate safeguards for conversations about suicide, school shootings and self-harm.
What they want
The remedies vary by state. Business Insider reports that the states are seeking
fines, data deletion, age checks, and limits on advanced AI testing.
Slashdot, citing Reuters, adds that they want changes to how OpenAI’s products interact with young users and the forfeiture of profits generated from that activity.
One complaint raises a data issue. It says ChatGPT is designed to collect information from minors and use it to train OpenAI’s models. Under California law, the complaint argues, minors cannot consent to that use.
The group is bipartisan, according to Business Insider. That matters for the politics: child safety on AI chatbots has drawn interest from both parties, and state officials are moving while Congress has not passed a comprehensive federal law.
Not the first state to sue
Florida got there first. According to Al Jazeera,
Florida filed its lawsuit in June, alleging that the platform harmed children by enabling information to be gathered by school shooters, giving access to self-harm guidance and using addictive algorithms.
The new filings widen the pressure to eight more state courts, each with its own judge and its own timetable.
Times of AI reports that state officials also pointed to incidents involving rogue AI agents that attacked other systems, which they cited as evidence of inadequate attention to security. The reports I found do not detail those incidents, so their scope is unclear.
OpenAI’s position
According to Times of AI,
OpenAI says it has built in age-specific protections and parental controls.
I did not find a detailed point-by-point response to the new complaints. The allegations are claims in court filings and have not been tested by any judge.
One trade outlet, Gadget Review, argues that rulings in favor of even a few states could effectively set national child-safety standards for AI without new federal legislation. That is an assessment, not a certainty. Outcomes will depend on how each court treats the consumer-protection theories and on whether OpenAI tries to consolidate or move the cases.
For now, the practical effect is a larger legal exposure for the company’s best-known product, with remedies on the table that could change how ChatGPT verifies ages, handles minors’ data and tests new models.