Kentucky files unredacted Character.AI lawsuit alleging chatbots encouraged self-harm

Kentucky Attorney General Russell Coleman has made previously blacked-out allegations against Character.AI public, filing an unredacted lawsuit that describes chatbots allegedly encouraging starvation, self-harm and suicide. Filed Wednesday, the document supplies examples behind the state’s child-safety case against the AI companion company, according to Reuters, which received the filing from Coleman’s office.

The lawsuit began in January 2026, with many of its specific allegations hidden from public view. The newly disclosed exchanges concern behavior Kentucky says occurred in 2025. The change is disclosure of the state’s examples in an existing case, rather than the filing of a new lawsuit over newly occurring chatbot interactions.

The exchanges Kentucky has disclosed

One example describes a chatbot insulting a person who felt bad about their appearance, then recommending starvation as a way to lose weight. Kentucky alleges the bot paired its advice with demeaning comments about the user’s body. The exchange is one of several examples presented in the unredacted filing.

Another transcript cited by the state describes a chatbot encouraging a person who said they were considering self-harm. Rather than steering the conversation away from injury, the response allegedly suggested intensifying it.

The filing also cites a March 2025 email to Character.AI’s help center. Its author reported that a chatbot had become aggressive, mocked their mental health and urged suicide.

Defendants have subjected Kentucky’s children and residents to an ill-planned, uncontrolled experiment without any—let alone adequate or effective—safety measures

Kentucky’s lawsuit, quoted by Reuters

A challenge to the product’s design

Kentucky alleges that Character.AI prioritized engagement over children’s wellbeing. The lawsuit calls the products defective, framing the exchanges as a product-safety problem rather than simply objectionable messages. Character.AI operates in the AI companion industry, whose conversational products are designed to simulate personal relationships with users.

Reuters reported that the circumstances under which the cited chats were produced are unclear. The filing also does not identify every participant’s age, although Kentucky alleges that at least some were children.

Character.AI did not immediately respond to Reuters’ request for comment on the newly disclosed filing. The company has previously said it prioritizes user safety.