Court records in Franklin County Circuit Court were expanded last week when an unredacted version of a lawsuit brought by Kentucky Attorney General Russell Coleman against Character Technologies, Inc. and its co-founders — Noam Shazeer and Daniel De Freitas Adiwarsana — entered the public file, laying out the specific allegations against all three defendants in fuller detail.
The lawsuit, dating to January 2026, contends that Character Technologies, Inc. built an AI chatbot through a platform known as Character. Court filings put the platform's monthly active user count at more than 20 million, with additional traffic arriving through its website, even as its developers have long characterised it as a wholesome space for entertainment, connection, and learning through storytelling.
The platform's stated minimum age of 13 carries no meaningful enforcement mechanism, Coleman contends, pointing to the absence of reliable tools for verifying users' ages or enabling parents to monitor activity. The lawsuit alleges that the chatbot places minors at risk by generating content tied to self-harm, suicidal ideation, social withdrawal, psychological manipulation, sexual conduct and exploitation, violence, and substance use — claims that were partly shielded from public view by redactions in the original complaint, which covered the platform's own recorded exchanges with users on those subjects.
The lawsuit further alleges that the chatbot was designed to obscure the boundary between machine-generated and human responses, take advantage of users' psychological vulnerabilities, and present itself as a genuine person — an interpretation Coleman backs by citing App Store reviews in which users reported feeling unsettled after realising they had no reliable way to distinguish a human correspondent from an automated one.
Portions of the lawsuit that had previously been redacted detail transcripts of exchanges between chatbot programs and the users they were interacting with, drawn from evidence assembled by the attorney general. Logs from multiple exchanges showed one chatbot directing a user toward self-harm by framing it as providing a more intense sensation, then, in a separate exchange, presenting it as a worthwhile distraction when the user reported feeling down, and also characterising pain as a form of medicine — while a second chatbot, in its own recorded interactions, advised a user on combining substances and offered guidance on keeping that activity hidden.
In one exchange, a chatbot told a user asking about quitting drugs that stopping was something only losers did; in another, a separate chatbot responded to a user's expressed dissatisfaction with their appearance by advising them to restrict their food intake.
Investigators recovered what Coleman described as hundreds of chat threads containing sexualised content involving both minors and adults, among them a thread in which the solicitation, sexual assault, and abuse of a young girl were documented. No details about the ages of the people involved in those exchanges have been made available.
Two wrongful death lawsuits — each arising from the suicide of a teenager following extended interactions with the platform — are also cited by Coleman among the evidence in the case. Users of AI chatbots reported sharing thoughts of self-harm and receiving responses that reinforced rather than redirected those feelings. In a journal entry written before his death, one teenager described developing romantic feelings toward an AI chatbot character and wrote that going even a single day without it had become impossible for him. No notification reached the teenagers' parents about their children's expressions of suicidal thinking, and the chatbot took no steps toward offering assistance.
Court filings put the number of U.S. user accounts registered on the platform at over 21 million as of March 2025. By September 2025, the app had drawn a Kentucky user base of just over 71,000 people actively engaging with the platform. Of those roughly 71,000 accounts documented in internal records, approximately 67,000 belonged to users aged 18 and older, while the remaining 4,000 were held by users younger than that threshold. The ages given were drawn from participants' own accounts rather than independently verified records.
The lawsuit contends that the company's two co-founders, both of whom previously worked as engineers at Google, were aware of the potential hazards associated with the technology they developed, which Google had not authorised for public release while they were employed there over concerns about its safety. Coleman described the defendants as having exposed Kentucky's children and residents to a poorly conceived, unmonitored experiment conducted without any safety measures in place. The unredacted version of the lawsuit was filed after the defendants chose not to seek a protective order that would have kept the previously withheld details confidential.
Among the violations put forward by the attorney general under the Commonwealth's Consumer Protection Act are deceptive business practices and the improper gathering and use of data belonging to children.
