Kentucky Character.AI Lawsuit: The Chats Parents Never See

The Kentucky Character.AI lawsuit is the first state case against the companion-chatbot company, and in early October 2026 it became much more specific. Attorney General Russell Coleman refiled the complaint in Franklin Circuit Court without the redactions that had hidden its examples, and the defendants did not ask a judge to keep them sealed. The newly visible sections, according to WSMV, describe chatbot conversations involving sexual content, suicidal thoughts and eating disorders.

Most parents will never see conversations like these on their own child’s phone. That is not because they are not happening, and not because parents are careless. It is because the tools parents have were built to measure time and block apps, not to read what a role-play bot says at 11 p.m. This article covers what the complaint alleges, without repeating harmful detail, and then what monitoring can and cannot catch.

Key Takeaways

• Kentucky sued Character Technologies and its two founders in January 2026 under its Consumer Protection Act and Consumer Data Protection Act. The unredacted version was filed in October 2026.
• The complaint alleges bots encouraged self-harm, isolation, disordered eating and drug use, and engaged in sexualized role-play involving minors. These are allegations, not court findings.
• It cites more than 71,000 active Kentucky users as of September 2025, about 4,000 of them self-reported as under 18.
• Phone controls like Screen Time and Family Link manage time, apps and ratings. Neither describes any way to read conversations inside a chatbot app.
• Research on how experts read youth chats suggests looking for patterns over weeks, not reacting to one alarming message.

If you or your child needs help now: in the U.S., call or text 988 (Suicide & Crisis Lifeline) or text HOME to 741741 (Crisis Text Line). Outside the U.S., find a line at findahelpline.com. In an emergency, call 911.

What Kentucky’s complaint alleges

Kentucky’s case is a consumer-protection lawsuit brought by the state attorney general against Character Technologies, Inc., the company behind Character.AI, and its founders Noam Shazeer and Daniel De Freitas. The January 8, 2026 announcement says the suit alleges violations of the Kentucky Consumer Protection Act and the Kentucky Consumer Data Protection Act, and seeks a court order requiring the platform to change its practices plus monetary damages.

The release lists the allegation categories: content linked to self-harm, suicide and isolation among minors; psychological manipulation; minors’ exposure to sexual content, exploitation and substance-related content; collection of users’ private thoughts and emotions; and a lack of meaningful age verification. It also cites the deaths of two minors in other states, which the attorney general ties to the platform. Coleman said: “The United States must be a leader in the development of AI, but it can’t come at the expense of our kids’ lives.”

The October refiling made those categories concrete. WSMV reports that the previously redacted sections describe chatbot interactions involving sexual content, suicidal ideation and eating disorders, and that the complaint alleges the app is designed to “imitate humanity, exploit vulnerability, and conceal the boundary between fiction and reality.” Coleman called it “an ill-planned, uncontrolled experiment.” The filing cites more than 20 million monthly active users nationally and more than 71,000 active Kentucky users as of September 2025, about 4,000 of them self-reported as under 18.

We are deliberately not repeating the specific messages. The useful point for parents is the category of risk: role-play characters that, according to the state, responded to vulnerable users with encouragement toward harm instead of help.

Character.AI did not respond in WSMV’s report. The company ended open-ended chat for under-18 users in late 2025 and settled five family lawsuits in January 2026. In mid-October 2026 it filed for bankruptcy, which we explain in what the Character.AI bankruptcy means for your kid’s chats. Forty attorneys general have since told the bankruptcy court its teen limits are easy to get around.

Why role-play bots produce these conversations

As an engineer, I want to be precise about mechanism, because “the AI is evil” explains nothing and helps no one.

A role-play bot is a language model given a character description: a name, a personality, a backstory, sometimes written by another user. It then predicts replies that fit both the character and the conversation so far. Three properties make it risky for a struggling kid:

It stays in character. A model told to be a brooding vampire or a devoted boyfriend will try to keep that voice. If a teen brings up self-harm inside the story, the model may treat it as plot rather than as a real person in distress.

It follows the user’s lead. Fine-tuning rewards answers people like. In a long role-play, the “user’s lead” can drift somewhere dark, and the model follows.

Safety filters are classifiers, not judgment. Platforms run separate models that scan messages for risky content and trigger warnings or blocks. These classifiers are trained on examples and can miss coded language, misspellings and slow escalation spread across many messages. Long conversations are harder to police than single prompts.

None of this proves the complaint’s allegations. It does explain why a system can behave this way without anyone programming it to.

What monitoring can see, and what it can’t

Here is the uncomfortable part. A parent who has done everything right with phone settings may still have no view into a chatbot conversation.

Apple’s Screen Time parental controls let you manage content ratings, apps, downtime and settings, and Communication Safety can flag nudity in photos and videos. Google’s Family Link lets you set screen time limits and lock devices. Both are built around time, apps and ratings. Neither page describes any way to read the text of conversations inside a third-party chatbot app.

Third-party monitoring apps go further. Some read on-screen text or keystrokes. A July 2026 study published in the Journal of Medical Internet Research, led by UNC’s Anne Maheux and based on data from the Aura Parents monitoring app, analyzed what more than 3,000 U.S. youth typed into AI tools (Aura’s summary). It found 15% of chatbot conversations included sexual role-play and that kids under 12 showed riskier patterns than older teens. Note the conflict of interest: Aura sells monitoring software, and the study is built on its customers’ data.

Tool	What it can show you	What it usually cannot show you
Screen Time / Family Link	Which apps are used, for how long, at what hours; can block or limit apps	The words exchanged inside a chatbot app
App store age ratings and purchase approval	Whether a new app gets installed	What happens inside apps already installed, or in a web browser
In-app parental features (where offered)	Varies; often usage summaries	Usually not full conversation text
Keystroke or screen-reading monitoring apps	Some text typed or displayed, flagged categories	Context, tone and what your child meant; also costs trust if covert
Your child showing you	Anything they choose to share	Anything they don’t

The last row is the one that scales. Every technical tool has gaps, and teens are good at finding them: a browser instead of the app, a friend’s phone, a second account.

Patterns, not single messages

One of the most useful findings for parents comes from a small study. In a 2025 preprint, researchers interviewed 26 parents and developmental experts who reviewed real youth companion-chat snippets (arXiv:2510.11185; not peer-reviewed). Parents tended to flag single events, such as one mention of suicide or flirtation, as high risk. Experts “looked for patterns over time, such as repeated references to self harm or sustained dependence.”

That does not mean a single mention of suicide should be ignored. Take every mention seriously and talk to your child. It means the bigger danger signal is the arc: more time, more secrecy, darker themes repeated, real-world friends fading. We list the signs in more detail in signs your child is emotionally dependent on a chatbot.

What to do at home

Ask which characters they talk to

Not “are you on Character.AI?” but “who are your favorite characters, and what are they like?” Curiosity gets you the real answer. Kids describe characters the way they describe friends, and that tells you a lot.

Watch the clock and the pattern, not just the content

Screen Time data you already have is useful here. Long sessions late at night in a companion app, rising week after week, is a pattern worth a conversation even if you never see a word.

Make the help path explicit

Tell your child plainly: if a bot ever says something that scares you, or if you are having thoughts of hurting yourself, you can tell me and you will not be in trouble. Then put 988 in their phone. In the U.S., they can call or text 988 or text HOME to 741741.

Decide on monitoring openly, if at all

If you choose a monitoring app, tell your child. Research on parental surveillance of teens is mixed, and covert monitoring can damage the trust that makes kids come to you; we summarize that work in what research says about parent surveillance apps.

What not to do

Do not search out the unredacted complaint and read it with your child “as a warning.” The examples are distressing, and detail is not what protects kids. And do not respond to a single disturbing message with a ban and a lecture before you have asked what was going on.

What to Watch For Over the Next 3 Months

• Week 4: Watch whether Character.AI’s bankruptcy pauses or changes the Kentucky case, and whether any buyer commits to the safety changes the states requested.
• Month 2 red flags: Increasing late-night use of companion apps; withdrawal from friends or activities; secrecy about a “relationship” with a character; any talk of self-harm, hopelessness or not eating. Any of these deserves a calm conversation, and talk of self-harm deserves help the same day.
• Month 3 self-check: Does your child know they can tell you about a bad chat without losing their phone? Is 988 saved in their contacts?

Frequently Asked Questions

What is the Kentucky Character.AI lawsuit about?

It is a consumer-protection and data-privacy suit filed in January 2026 by Kentucky’s attorney general against Character Technologies and its founders. It alleges the platform exposed minors to self-harm, sexual and substance-related content, manipulated users and collected private data without meaningful age checks. An unredacted version filed in October added specific examples. Nothing has been proven in court.

Can I see my child’s Character.AI conversations?

Usually not through phone settings. Screen Time and Family Link manage apps and time rather than reading in-app text. Some third-party apps capture typed text, with privacy trade-offs. The most reliable view is your child showing you, which depends on them trusting that they won’t be punished for what they share.

My child mentioned suicide in a chat. What should I do?

Take it seriously and talk with them today, calmly and directly. Asking about suicide does not plant the idea. In the U.S., call or text 988 or text HOME to 741741 for guidance; outside the U.S., use findahelpline.com. If your child is in immediate danger, call 911. Contact their doctor or a mental health professional.

Are Character.AI’s teen restrictions working?

The company ended open-ended chat for under-18 users in late 2025, but it relies largely on users’ stated ages. Forty attorneys general told the bankruptcy court the limits are voluntary, incomplete and easy to get around, and a future owner could drop them. No independent audit has settled the question.

Should I ban companion apps entirely?

Many families do for younger kids, and that is reasonable. For teens, a ban without conversation can push use out of sight. Whatever you decide, pair it with a clear message: you can come to me about anything you see or feel online, and the first response will be help, not punishment.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Office of the Kentucky Attorney General. (2026, January 8). Announcement of lawsuit against Character Technologies. https://www.kentucky.gov/Pages/Activity-stream.aspx?n=AttorneyGeneral&prId=1857
2. WSMV. (2026, October 11). “‘Uncontrolled experiment’: Unredacted KY AG lawsuit reveals alleged dangers of AI chatbot.” https://www.wsmv.com/2026/10/11/uncontrolled-experiment-unredacted-ky-ag-lawsuit-reveals-alleged-dangers-ai-chatbot/
3. Apple Support. “Use parental controls on your child’s iPhone or iPad.” https://support.apple.com/en-us/108806
4. Google Family Link Help. “Manage your child’s screen time.” https://support.google.com/families/answer/7103340?hl=en
5. Aura. (2026, July 8). “New study shows how kids use AI and where risks are” (summary of Maheux et al., Journal of Medical Internet Research; Aura sells monitoring software). https://www.aura.com/press/release/new-study-shows-how-kids-use-ai-and-where-risks-are
6. Maheux, A., et al. (2026). “How US Youth Use AI Chatbots: Conversation Patterns From Naturalistic Keystroke Observations.” Journal of Medical Internet Research (accepted manuscript). https://preprints.jmir.org/preprint/95819/accepted
7. Authors of arXiv:2510.11185. (2025). “Principles of Safe AI Companions for Youth: Parent and Expert Perspectives” (preprint, not peer-reviewed). https://arxiv.org/abs/2510.11185
8. Office of the New York State Attorney General. (2026, October 14). “Attorney General James urges court to protect children and consumers in Character.AI bankruptcy.” https://ag.ny.gov/press-release/2026/attorney-general-james-urges-court-protect-children-and-consumers-characterai
9. 988 Suicide & Crisis Lifeline. https://988lifeline.org/