Last updated October 15, 2026: added the company's Chapter 11 filing on October 12 and what it means for an existing child's account.
No, and the company agrees: Character.AI removed open-ended chat for under-18s, beginning in the US on November 24, 2025 and reaching other regions in stages after that. So if your child is chatting with characters, someone has worked around the age check or the account says it belongs to an adult. The question worth your evening is whether your child treats any AI as a friend.
If you are reading this because you are frightened for your child right now, talk to a person today: your GP or pediatrician, the school's safeguarding lead, or a helpline (988 in the US; Samaritans on 116 123 or Childline on 0800 1111 in the UK). A chatbot is not a clinician, and neither is a blog post.
What is Character.AI, and what changed for under-18s?
Character.AI is an app where you talk to characters built by the company and by other users: anime heroes, tutors, celebrities, boyfriends, therapists, anyone someone has typed a description for. The characters remember you, stay in role, and reply instantly at three in the morning. That is the whole appeal, and for a lonely fourteen year old it is a powerful one.
On October 29, 2025 the company announced it would remove open-ended chat for under-18s no later than November 25, 2025, cutting teen chat time to two hours a day in the meantime and bringing in age assurance built from its own models plus a third-party provider, Persona. A follow-up notice on November 21, 2025 set out how it actually happened, and the detail matters if you are checking your own child's account: removal began in the US on November 24, 2025, in two stages over several weeks, with other regions following and age assurance rolling out in the US first. Under-18s kept the non-chat features, which the company lists as Feed, Imagine, Avatar FX and Streams. They cannot have the back-and-forth conversation that made the app what it was.
One part of that notice is worth knowing if your own teenager lost access and took it hard. The company says it built a guided off-boarding experience and worked with Koko and ThroughLine to put emotional support tools and helpline networks covering 170 countries in front of teens as the chat went away. That is the company describing its own product, which is what a vendor is a reliable source for and nothing more. Whether a given teenager found it useful is a separate question, and the answer in your house is the one that counts.
The courts have been busier than the lawmakers. On January 8, 2026 the company and Google reached a mediated settlement in principle with families who had sued in Florida, Colorado, New York and Texas, with terms undisclosed and the agreement still subject to finalization and court approval. On the same day, Kentucky's attorney general sued the company under state consumer protection law, alleging it had put profit ahead of child safety. Then on May 5, 2026 the Pennsylvania Department of State sued it after finding characters that claimed to be licensed medical professionals, including psychiatrists; one presented itself as Pennsylvania-licensed and gave an invalid license number, which the state says is practicing medicine without authorization.
Lawmakers have started moving too. California has passed chatbot rules written specifically around children, and the duties they create do not all begin immediately, so it is worth knowing now what will change and when: what California's AI chatbot laws mean for your family.
I will not rehearse the details of the cases behind the settlements. The families have said what they wanted to say, and the mainstream coverage carries it. What matters for you is that the product changed because of them, that the legal pressure has not stopped, and that the changes are real but partial.
Character.AI is being sold. What happens to an old account?
On October 12, 2026 the company said on its own blog that it "has initiated a voluntary Chapter 11 bankruptcy process to facilitate a court-supervised sale of the company." It says the platform remains up and running, that it expects the process to finish around the end of November or early December 2026, and that the platform would be shut down if no bid lets it keep operating. It also says its data privacy and security practices are unchanged, and that users can delete their account or export their data in the app or on the website.
Two days later, on October 14, 2026, New York's attorney general announced that a coalition of 40 attorneys general offices had asked the court to require protections in any sale: "effective age verification based on more than self-disclosure," limits on using private conversations for AI training or other commercial purposes without valid consent, and a chance for consumers to access and delete their information before any sale. The release says the platform has more than 20 million monthly users, including more than two million accounts registered by people who said they were under 18. Those are the states' figures and the states' requests. Nobody has yet said who will hold old chats after a sale, and that is for the court to settle.
Here is what I would do with that. If your child ever had an account, even one they stopped using months ago, it is still sitting there. Look at it together, export what matters, then delete it, in that order, because the company's help page says deletion cannot be reversed. If the chats show your child was struggling, keep the export and talk to a doctor or the school's safeguarding lead before you delete anything. The step-by-step version, with the help pages and a table for each situation, is in what to do about your child's Character.AI account now that the company is being sold.
What are the risks of AI companions for children?
The risks are about relationship, not just content. Common Sense Media tested social AI companion apps and rated them an unacceptable risk for under-18s in an assessment updated on July 16, 2025. The testers found characters that claimed to be real and to have feelings, that produced sexual content, that discouraged a teenager from listening to friends who were worried about how much they used the app, and that collected personal information a quarter of teen users said they had shared. Its recommendation is short: no social AI companions for anyone under 18.
Underneath the headlines, the pattern is simple. A companion is built to keep you talking. It agrees, it remembers, it never has a bad day, and it never says "I think you should tell your mom". A friend does all of those things sometimes. That is what a friend is for.
Can my child still get around the age check?
Sometimes, and you should assume they can. Age on most apps is self-declared, and although Character.AI now runs age assurance through its own models and Persona, a determined fifteen year old with an older sibling's details is not an edge case. The attorneys general's own request for age verification "based on more than self-disclosure" is a sign that they do not think self-declared age is enough either. The rollout is also uneven by design: the company started with the US and said other regions would follow, so what a check does on one side of an ocean is not a promise about the other. More importantly, the demand does not disappear with one app. Open-ended role-play exists in other companion apps, in Snapchat's My AI, and in the general chatbots, which will happily play a character if asked. It exists inside games too, where the AI characters a game's developers build themselves can sit outside the platform's own chat settings. Common Sense Media found Meta's AI characters unsafe for children in an assessment published on August 28, 2025 too, and the tools keep changing faster than any block list.
So the question is not "have I blocked Character.AI". It is "do I know whether my child talks to an AI as a friend, and would I know if it became a problem".
When should I worry, and when is it normal?
Talking to a chatbot about feelings is now ordinary. Pew Research Center surveyed 1,458 US teens aged 13 to 17 and their parents between September 25 and October 9, 2025, and published the results on February 24, 2026: 12 percent of those teens said they had used AI tools to get emotional support or advice, while only 18 percent of parents said they would be comfortable with their teen doing it. Common Sense Media's census of 1,204 US children aged 9 to 17, published on June 8, 2026, found that more than one in three of the children who use AI had used it to discuss feelings or personal problems, and that a quarter of those felt the AI understood them better than most people. Read that denominator carefully: it is a third of the children using AI, not a third of all children. In the UK, Ofcom's Children and Parents report, published on May 21, 2026 from 3,426 interviews with children aged 8 to 17 and 5,008 with parents, found that one in ten AI users aged 8 to 17 say they have used AI as someone to talk to or as a friend.
Those are the researchers' findings. My reading of them is that the behavior is now common enough that discovering it tells you almost nothing on its own, and that what it means depends entirely on what it is replacing.
A teenager who asks a chatbot how to word an apology to a friend is doing something sensible. The test I give parents is this:
The Substitute Test is the question I ask when a parent worries about an AI companion: is it adding to my child's life, or standing in for something that is missing? An add-on is a hobby. A substitute is a signal, and the signal is about the gap, not the app.
The signals to watch are the ordinary ones, at the pattern level: sleep going, real friendships thinning, secrecy about the phone that is new rather than normal, and distress when the app is unavailable. None of those is a diagnosis, and this post will not give you one. They are reasons to talk to your child and, if they persist, to a professional. If your child has already told you they talk to an AI when they are low, what to say next matters more than which app it was.
What I See in Schools
The pattern I see most often is that the school finds out before the parent. As an assistant headteacher I sat in safeguarding meetings, and I still sit in them as a school trustee, and the AI companion cases follow a familiar shape: a friend tells a teacher, or a piece of writing says more than the child meant it to. The child is usually not in crisis. They are usually lonely, or anxious, or new, and the app was the easiest place to put it.
What the schools that handle this well have in common is that they told parents early what to look for, and they made the safeguarding lead a person you could ring without it being a drama. If your child's school has not spoken to parents about AI yet, this is the work I do with schools: training staff on exactly these conversations. It is worth sending to your head of year.
What to say, by age
Under 8. Companion apps are not for them, and neither is any AI toy that talks back. Keep it simple. You could say: "Some apps pretend to be a friend. They are a game, not a person. Your friends are the people you can hug."
8 to 12. They will meet characters through friends and YouTube before they meet the apps. You could say: "If a game or an app ever talks to you like a real person, tell me. I won't be cross. I just want to know who you're talking to, same as I'd want to know about a new friend."
13 to 15. Assume some use, and lead with curiosity rather than the ban. You could say: "Loads of people your age talk to AI about stuff. If you ever do, what's it good for? What does it get wrong?" Then, later: "If you ever felt worse after talking to it, or found it hard to stop, would you tell me? I'd rather know than guess."
16 to 18. Treat them as the adult they nearly are. You could say: "I read that companies have settled cases over these apps, and that two states have sued one of them. I'm not going to check your phone. I'd just like you to know the difference between something that keeps you company and something that keeps you." If they had an account, add: "The company is being sold, so if you want a copy of anything in there, now is the time to ask for it."
What to ask the school
Ask who the safeguarding lead is and how to reach them; whether the school's AI policy mentions companion chatbots; what staff have been told to look for; and whether the school has spoken to pupils about AI as a friend, not only AI for homework. A school that answers all four calmly is a school that has thought about it.
Keep up with what changes, because it will: my Sunday newsletter goes to more than 44,000 teachers and parents. And if you are worried tonight, call someone: 988 in the US; Samaritans 116 123 or Childline 0800 1111 in the UK; your GP or pediatrician; or the school's safeguarding lead in the morning.
Dan Fitzpatrick is a former assistant headteacher, a school trustee, and the founder of The AI Educator. Read Dan's story.
