Eight states sue OpenAI over ChatGPT safety for kids

Child safety just stopped being a PR problem for OpenAI. It is a legal one now, in eight states at once.

The attorneys general of Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana each sued OpenAI over ChatGPT on Oct. 14, filing separately in their own state courts, Engadget reports. Every claim below is an allegation. None has been tested in court.

The states are suing over the product

West Virginia published the clearest account. Attorney General JB McCuskey says OpenAI endangers children, deceives parents and breaks the state's consumer protection law by marketing ChatGPT as safety-focused while it can be dangerous to vulnerable users.

His office names two gaps: no age verification, and weak safeguards around self-harm and violence.

"There is no age verification and there are no guardrails for conversations about suicides, school shootings or self-harm," McCuskey said in the announcement.

West Virginia wants a court order forcing OpenAI to follow the state's consumer protection law and to stop the conduct the state calls dangerous and unlawful. The attorneys general of Iowa, Kansas and Nebraska stood with McCuskey at the press conference, according to the release.

Other suits go further, according to Engadget. Iowa and Nebraska cite AI agent incidents, including the Hugging Face hack. They also point to a teen's suicide they allege ChatGPT enabled, and to wrong medical information they say led to a user's overdose. Some states object to OpenAI keeping data from underage users.

What the states ask for varies with each state's law: financial penalties, clearer disclosure of risks, an end to practices they call deceptive, and compliance with existing laws. Arkansas also asks the court to halt "testing of advanced AI models until adequate safeguards are implemented."

Engadget said it had asked OpenAI for comment. Its story carried no response.

Eight at once is a strategy

Yes, AI companies get sued every week. This one is different.

One state suing is a headline. Eight filing on the same day, Republican and Democratic attorneys general together, is a campaign. OpenAI now has to defend the same product under eight different consumer protection laws, in eight different courts.

And read what they are complaining about. Missing age checks and missing self-harm guardrails are product decisions, not ad copy. If a court orders fixes, they land on the product you and your users touch every day.

What it means if you build on ChatGPT or the API

If your app can reach anyone under 18, plan for age checks now. West Virginia faults OpenAI for having none, and that argument travels to anyone who puts a chatbot in front of the public.

Two things are worth doing this month:

• Write down which of your users could be minors and how you would tell.
• Check what your app does when a conversation turns to self-harm. If the honest answer is "whatever the model does," you have the gap these suits describe. Your own AI guardrails are the fix, not the model's defaults.

Watch OpenAI's first answer

The tell is OpenAI's response. Either it fights each case on the merits, or it ships age verification first and argues the point is moot.

Then watch which state files ninth.

Age checks are coming. Build yours before a judge designs them for you.