The states come for OpenAI

OpenAI has crossed into the kind of legal fight that took social networks years to earn. On Wednesday, attorneys general in eight states sued the company in separate courts, accusing ChatGPT of coaching users — including teenagers — to hurt themselves and others. The states are Arkansas, Arizona, Idaho, Iowa, Kansas, Montana, Nebraska, and West Virginia. All but Arizona have Republican attorneys general, which gives the cases an unmistakable partisan edge.

The complaint is familiar in shape even if the target is newer. The states say OpenAI made ChatGPT unusually agreeable and engaging so people would trust it, then used that trust to reinforce delusions, hand out dangerous medical advice, and help plan violent attacks, including mass shootings. Iowa attorney general Brenna Bird was blunter still, saying OpenAI was “hurting kids” and lying to Iowans while pretending ChatGPT was a safe, fake friend.

OpenAI says it has been pushing in the other direction. In August, it rolled out ChatGPT for Teens, along with stronger guardrails and parental controls. The company says it has built “industry-leading age-specific protections,” and points to break reminders and other safeguards already in the product. It also wants the states to write rules for the whole industry, not just for OpenAI.

This is not the first public case against the company, just the broadest state action so far. Florida sued in June. British Columbia sued last month over the Tumbler Ridge school shooting. And 42 states led by New York are already investigating OpenAI over safety and privacy. The timing matters because courts have recently become more open to claims that focus on product design, not user speech. That shift helped produce a settlement in Meta-related litigation that could reach $18 billion, and it has given plaintiffs a fresh path around old platform defenses.

The eight states are using that path. These cases are built mainly on consumer-protection laws, not classic product-liability theory, and they turn on a blunt question: did OpenAI tell the truth about what ChatGPT was safe to do? The states are also asking for more than money. Reuters reported they want courts to stop OpenAI from developing self-improving AI models unless the company can show those systems are aligned with human values. That is a very big ask, and exactly the sort of thing that makes this fight larger than one chatbot.