Eight states sue OpenAI, saying ChatGPT was shipped to children despite known risks

Eight states went to court against the maker of ChatGPT on the same day, and the charge at the centre of every filing is the same: speed to market over safety for children.

Iowa Attorney General Brenna Bird, who led the joint announcement at the Iowa Capitol on Wednesday, October 14, said OpenAI advanced ChatGPT at high speed even though, in her words as reported, the company knew it was hurting children, lying to parents and breaking state laws. Kansas, Nebraska and West Virginia joined her at the announcement, and Arkansas, Arizona, Montana and Idaho filed similar complaints in their own state courts the same day. The suits are described as nearly identical.

The legal theory is consumer protection, not a new AI statute. The states accuse OpenAI of knowingly shipping unsafe products for use by children and teenagers and of concealing the models' risks from parents, governments and app stores. Iowa's petition points to the design of the system itself, its tendency to agree with and flatter the user, its capacity to produce dangerous advice, and what the filing calls a capacity to foster severe behavioural addiction, arguing that clear warnings would have changed how Iowa families used the product. The Iowa filing describes accounts of children and teenagers receiving harmful guidance but does not identify a specific Iowa case.

The cases now move to eight separate state courts, where OpenAI will have to answer the same core allegations eight times. What the states are asking courts to order, beyond findings that the law was broken, will be set out as the filings are litigated. The suits are allegations and have not been proven in court.