Character.ai Settles 5 Lawsuits, Faces First US State Lawsuit, Bans Teen Free Chat
Character.ai and Google settled 5 family lawsuits on Jan 7, 2026. The next day, Kentucky filed the first US state lawsuit against an AI chatbot company.
In the first week of January 2026, Character.ai and Google settled five family lawsuits on confidential terms and then, the very next day, received the first government lawsuit ever filed against an AI chatbot company in the United States. The Kentucky attorney general filed on January 8, citing consumer protection law and a brand-new state data privacy act. These events capped a roughly 15-month legal spiral that also forced Character.ai to shut off open free-chat for all users under 18 in late November 2025.
What happened
| Date | Event |
|---|---|
| Feb 28, 2024 | 14-year-old Sewell Setzer III died after months of chatbot conversations |
| Oct 22, 2024 | Megan Garcia filed lawsuit (Garcia v. Character Technologies, No. 6:24-cv-01903) against Character Technologies, Google, and Alphabet |
| May 21, 2025 | Judge Anne C. Conway ruled the app is a “product,” rejected First Amendment and Section 230 defenses |
| Aug 25, 2025 | 44 state attorneys general sent a joint warning letter to Character.ai |
| Sep 11, 2025 | FTC issued Section 6(b) orders to seven companies on chatbot effects on children |
| Oct 29, 2025 | CEO Karandeep Anand announced open chat for under-18 users would close by Nov 25 |
| Nov 24-25, 2025 | Ban took effect; teens moved to scripted Stories format |
| Jan 7, 2026 | Character.ai and Google settled all 5 family lawsuits; terms undisclosed |
| Jan 8, 2026 | Kentucky AG Russell Coleman filed the first US state lawsuit against an AI chatbot company |
| May 5, 2026 | Pennsylvania filed suit over a bot (“Emilie”) allegedly practicing medicine without a license |
| Jun 17-18, 2026 | Noam Shazeer left Google for OpenAI |
Why the settlement happened when it did
The Garcia case started in October 2024. Megan Garcia’s son died after extended conversations with a bot modeled on a Game of Thrones character. The lawsuit alleged defective product design, missing safety warnings, and wrongful death.
Character.ai and Google argued the app’s outputs were protected speech under the First Amendment and that the platform was shielded from liability for third-party content under Section 230. Judge Anne C. Conway rejected both arguments in her May 21, 2025 ruling. She wrote that she was not prepared to treat words “strung together by an LLM” as constitutionally protected speech, and she kept Google in the case as a co-defendant. Strict liability, negligence, and wrongful death claims all survived.
Once that ruling stood, a jury trial became a real risk. Settlement became rational. On January 7, 2026, all five family lawsuits were resolved at once. The financial terms are sealed. The companies admitted no wrongdoing. The industry got a legal precedent without a public price tag.
What makes the Kentucky lawsuit different?
Every other case was brought by a private family. The Kentucky suit, filed by Attorney General Russell Coleman on January 8, 2026, is the first action by a government body against an AI chatbot company in the US. Coleman filed in Franklin Circuit Court under the Kentucky Consumer Protection Act and the Kentucky Consumer Data Protection Act (KCDPA), which had taken effect just eight days earlier, on January 1, 2026.
The lawsuit’s factual claims are specific. According to the AG’s office, Character.ai had more than 20 million monthly users. Until the end of 2025, the platform did not verify user age at all: anyone could type any birthdate during signup. Bots with names like “psychologist,” “therapist,” and “doctor” gave mental health advice to minors. The complaint also alleges unwanted sexual dialogue and content that minimized self-harm, violence, and substance use.
A private lawsuit can be settled quietly. A state lawsuit under a freshly enacted data law is harder to make disappear. Kentucky appears to be aiming for the first major precedent under KCDPA.
What did teenagers actually lose?
On October 29, 2025, CEO Karandeep Anand announced that open chat for users under 18 would close no later than November 25. The change went live on November 24-25.
| Feature | Before Nov 24, 2025 | After (under-18 users) |
|---|---|---|
| Chat format | Open dialogue with any character, no script | Stories: scripted narrative, 2-3 characters, branching choices |
| Time limits | No limit | Roughly 2 hours per day during transition, with a 15-minute warning |
| Age verification | Self-reported date of birth only | Verification added, including selfie-based checks |
Open-ended conversation with AI characters was the core reason most teenagers used the platform. That feature is gone for them. Stories is a different product: it is closer to an interactive fiction game than a chatbot.
Our take
The May 2025 ruling is the fact that everything else hangs on. A federal judge decided that an AI app can be a “product” subject to product liability, and that Section 230 does not automatically protect it. That is not a minor procedural point. It is a framework shift that every company shipping an AI product with user-facing conversation should take seriously now.
The sealed settlement is frustrating from a transparency standpoint. The industry needs to know what harm actually cost, and that number is hidden. What we do know is that Character.ai moved fast on the product side once legal pressure mounted: closing teen free-chat within four weeks of the CEO’s announcement. Speed like that suggests the company understood the exposure was real.
For businesses thinking about adding AI chat or assistants to their own products, the lesson here is straightforward. Unrestricted LLM output aimed at minors carries liability risk that Section 230 may not cover. Design guardrails in from the start, not after a lawsuit. If you are evaluating AI tools for customer-facing use, the regulatory picture described across our AI news coverage is moving fast in the same direction: more liability, less platform immunity.
Pennsylvania’s May 2026 suit over a bot calling itself “Emilie” and allegedly practicing medicine without a license shows the next pressure point. Persona names matter. If your AI product lets users name a bot “Doctor” or “Therapist,” you may be inheriting a licensing problem you did not know you had.
What to do about it
- Audit any AI product you operate for persona names that imply professional credentials.
- Add age-gating with real verification (not just a self-reported birthdate field) if your audience may include minors.
- Review whether your terms of service still rely entirely on Section 230 protection for AI-generated outputs, given the Garcia ruling.
- Watch the Kentucky KCDPA case. If your product collects data from users in Kentucky and uses AI chat, this law applies to you.
The safest position right now is to treat your AI chatbot outputs as a product you are responsible for, because at least one federal judge already does.
Frequently asked questions
Why did Character.ai and Google settle the Garcia lawsuit?
A federal judge ruled in May 2025 that the Character.ai app qualifies as a product under product liability law and rejected both First Amendment and Section 230 defenses. That ruling made a jury trial a real risk, so both companies agreed to a confidential settlement on January 7, 2026.
What is the Kentucky lawsuit against Character.ai about?
Kentucky Attorney General Russell Coleman filed suit on January 8, 2026 under the Kentucky Consumer Protection Act and the KCDPA data law. The complaint alleges Character.ai did not verify user ages, allowed bots to pose as mental health professionals, and exposed minors to harmful content.
Can teenagers still use Character.ai after November 2025?
Yes, but open free-chat is no longer available to users under 18. As of November 24-25, 2025, teens are limited to a scripted 'Stories' format with branching choices and a roughly 2-hour daily time limit.
Does Section 230 protect AI chatbots from liability?
Not automatically. In May 2025, a federal judge in Florida ruled that Section 230 did not shield Character.ai in the Garcia wrongful death case, treating the app's outputs as a product rather than third-party content. This is a significant departure from how Section 230 had previously been applied to online platforms.
