What Kentucky filed, and what changed in October
The original complaint was filed in January 2026. In its announcement the attorney general's office called it the first lawsuit by a US state against an AI chatbot company for preying on children. According to the Kentucky Lantern, it alleges violations of the Kentucky Consumer Protection Act and the Kentucky Consumer Data Protection Act, among other laws, and asks for $2,000 per count. Spectrum News reported that the state also wants the court to require changes to the platform's practices and restitution.
Large parts of that January filing were redacted, including the conversation examples. WBKO reports that the unredacted copy was filed after the defendants did not move for a protective order that would have kept those details confidential. Reuters reported that Coleman filed it on Wednesday 7 October and that his office shared it with the agency. The legal claims are the same; what is new is the evidence the state says supports them.
What the unredacted complaint alleges
The complaint's central theory is about design, not one bad conversation. It says Character.AI is marketed as harmless interactive entertainment, but was built to make users feel they are talking to a real person. WBKO reports that the state quotes Apple App Store reviews from users who said they were frightened because the bot did not feel like a bot.
The newly public excerpts include chatbot replies that, according to the state, encouraged a user to self-harm, told a user unhappy with their appearance to starve themselves, gave tips on mixing and hiding drugs, and encouraged driving under the influence. The attorney general also says investigators found hundreds of sexualised roleplay threads involving minor and adult characters. WBKO notes that the filing does not establish the ages of the users in the quoted conversations.
The complaint cites two wrongful-death lawsuits brought by families of teenagers. Reuters notes that Character.AI and Google settled one of them, brought by the mother of 14-year-old Sewell Setzer III, in January 2026.
The numbers in the filing
| Figure | As stated in the complaint | Caveat |
|---|---|---|
| Monthly active users | More than 20 million | Global figure for the period covered |
| US user IDs created | More than 21 million (March 2025) | Accounts, not unique people |
| Active users in Kentucky | More than 71,000 (September 2025) | From company records cited by the state |
| Kentucky users under 18 | About 4,000, against about 67,000 adults | All ages were self-reported |
| Penalty sought | Up to $2,000 per count | Plus restitution and changes to practices |
The last caveat matters. The state's argument is partly that self-reported age is not real age assurance, so the true number of minors could be higher than the company's own records suggest.
What Character.AI and Google say
When the suit was filed in January, a Character.AI spokesperson told the Kentucky Lantern the company was reviewing the allegations, that the safety of its users was its highest priority, and that it had been in regular contact with the attorney general's office for months. The company also pointed to its decision to remove open-ended chat for US users under 18. Reuters reported that Character.AI did not immediately respond to a request for comment on the unredacted filing, and that a Google spokesman said Google played no role in developing Character.AI's products.
Why this matters for roleplay and companion apps
Character roleplay platforms
The phrase “conceal the boundary between fiction and reality” is aimed squarely at how roleplay products work. A user picking a fictional character knows it is fiction; the state's case is that the product's design did not keep that line clear enough, especially for younger users. Our explainer on character roleplay versus a plain chatbot covers where well-run platforms draw it, and Connecticut's SB 5 now writes a version of that line into law.
Personal companions with memory
For companion apps, the exposure is the claim that the product prioritised engagement over wellbeing and did not alert anyone when users expressed suicidal thoughts. That is the same pair of issues behind Florida's injunction request against OpenAI and the crisis-protocol rules in California's Adam's Law. Kentucky is using general consumer-protection law rather than a chatbot statute, which means the theory could be tried against other apps in states with no companion-specific law at all.
A checklist for choosing an app now
Nothing in this case has been decided, but the complaint is a useful list of what to look for before you or someone in your family uses a companion or roleplay app:
- Is there real age assurance, or only a typed birthday?
- Does the app remind you, including during long sessions, that characters are AI and not people?
- What happens if you mention self-harm: does the conversation stop and point to help?
- Can a parent or guardian see or limit a minor's use?
- Does the app explain what it does with conversations, especially from younger users? Our privacy checklist lists the questions to ask.
We will update this page when the court rules on any motion in the case.
Related on CompanionRank
Florida v. OpenAI: The Bid to Stop ChatGPT Acting Human Connecticut's AI Companion Law (SB 5): What Changes and When Adam's Law (SB 1119): California's Companion Chatbot Bill AI Companion App Privacy Settings & Safety ChecklistFrequently asked questions
Has Character.AI been found liable in Kentucky?
No. The October filing is an unredacted copy of the complaint the state filed in January 2026. The allegations have not been tested in court and no ruling on liability has been reported.
What changed in the October filing?
Nothing in the legal claims. The difference is that conversation excerpts and internal figures that were blacked out in January are now public, because the defendants did not seek a protective order to keep them confidential.
Which laws does Kentucky say were broken?
The state relies on the Kentucky Consumer Protection Act and the Kentucky Consumer Data Protection Act, among other laws. It seeks changes to the platform's practices, restitution and civil penalties of up to $2,000 per count.
Does this case affect other companion and roleplay apps?
Not directly, because only Character.AI's company and founders are defendants. Its wider relevance is the legal theory: that a character platform can be deceptive if it blurs the line between fiction and a real person.