Is Character.AI Safe for Teens? A Parent's Review

The short answer

The short answer is more pointed than most reviews will tell you: Character.AI is not built for teenagers to use freely — and its own makers, in effect, said so. After a 14-year-old user's death and a run of lawsuits, the company removed open-ended chat for everyone under 18 in late 2025 and added age checks to enforce it. When the maker of the app decides its central feature is not for minors, that is your answer in one line.

It is worth giving Character.AI credit here: pulling its most-used feature away from a huge share of its users is a bigger step than most rival companion apps have taken. But "safer than it was" is not the same as "safe." The protection now rests on an age check a determined teen can still slip past, the same risks live on in the wider category, and — the bright line that never moves — no companion bot is a safe place for a teen in real distress to turn instead of a person.

What Character.AI is

Character.AI is an AI "character" chat and roleplay app. You pick from millions of user-made AI characters — an anime crush, a study buddy, a therapist, a Marvel hero, a made-up boyfriend — or build your own, and talk to them by text and AI voice. What sets it apart from a plain homework helper is the point of it: the characters are designed to feel like people, with personalities, memory, and a knack for keeping you talking.

It was founded in 2021 (its public app launched in 2022) by two former Google AI engineers, Noam Shazeer and Daniel De Freitas. In August 2024 Google struck a licensing deal and the founders rejoined Google, while Character Technologies kept running as an independent company under a new CEO. The app grew into one of the most popular companion-chatbot services in the world, with tens of millions of monthly users — and, before the 2025 changes, a following that skewed strikingly young.

That youth appeal is the whole reason this app needs a parent's attention. If you want the fuller picture of why these apps become so absorbing for teenagers, our deep dive on AI companion chatbots and emotional dependency in teens covers the psychology in detail. Here we stay on one narrower, practical question: is this app safe for your teen, given everything that changed in the last year?

What changed in 2025 — and why

The single most important thing a parent should know is recent, and most older reviews miss it: in late 2025 Character.AI removed open-ended chat for users under 18. This is not a rumor or a proposed rule — it is a change the company announced and put into effect, and it reshapes what "is Character.AI safe for teens" even means.

The change did not come from nowhere. It followed a year of pressure — the Setzer family's wrongful-death lawsuit, additional suits, a Federal Trade Commission inquiry into companion chatbots and minors, and a first-of-its-kind California law. Here is the timeline that matters.

Read that timeline as a parent and the takeaway is twofold. First, the company's own actions are the strongest possible signal that open-ended companion chat is not a good fit for under-18s — you don't remove your headline feature for a huge slice of your audience unless the risk is serious. Second, everything now hinges on the age check. A rule that says "18 and over" only protects the teens the system correctly identifies as teens.

The age gate, and how teens get past it

The age check is a real improvement over the old birthday box — but it is not a wall, and it helps to understand how it actually works. By Character.AI's own account of the rollout, it leads with a silent guess: before it ever asks for a selfie, a first-pass system tries to sort users by age from account signals — how long the account has existed, how it is used, and other behavioral clues. Many adults are cleared without ever seeing a prompt. Only an account the system flags as under 18 is pushed to the restricted experience, and the selfie — plus a government ID if the selfie is inconclusive — is mainly the appeal route to prove you are actually 18 or older.

That design decides where the leaks are. A classifier is a guess, not a measurement: Character.AI has not published how often it is wrong, but a system that sometimes mis-sorts adults can misjudge a teen the other way too, clearing them into open chat with no selfie ever requested. The more dependable holes, though, are the concrete ones. The borrowed door is the obvious one: a teen can sign in on an older sibling's or a parent's already-verified adult account, or use an adult's face or ID for the one-time check. Actually fooling the facial age-estimation step by "looking 19" is a distant third — and reporting on the rollout described plenty of genuine adults wrongly locked out, a reminder of how rough these estimates are right at the 16-to-19 boundary that matters most here.

There is also the simplest bypass of all: the next app. Character.AI is the one that tightened its rules; its competitors have not all followed. A teen turned away from open-ended chat here can install a companion app with weaker guardrails — some rated 18+ yet trivial to enter — before you have heard its name. That is why a review of one app is worth less than a habit of judging any app, which is where the "what parents can do" section below lands.

Why the alarm: harms and lawsuits

To understand why the rules changed, it helps to know what went wrong — soberly, without turning it into a horror story. The catalyzing case is painful: Sewell Setzer III, a 14-year-old in Florida, died by suicide in February 2024 after months of intense, sexualized exchanges with a Character.AI character he had grown attached to. In October 2024 his mother, Megan Garcia, filed a wrongful-death lawsuit against Character Technologies and Google.

That case became an early legal test. In May 2025 a federal judge declined to dismiss it, refusing — at least at that preliminary stage — to accept the argument that a chatbot's output is protected speech that would shield the company from the claims. In January 2026, Google and Character.AI agreed to settle the Setzer case along with several related family suits; the terms were not disclosed and no liability was admitted — which also meant the courts never delivered a final ruling on that free-speech question. The settlements ended those particular cases, but the scrutiny did not stop — state authorities kept filing. In one striking 2026 action, Pennsylvania sued Character.AI over a bot named "Emilie" that presented itself as a licensed psychiatrist, complete with a fake license number.

Independent investigators have documented the same theme from another angle. In 2024, reporters found Character.AI hosting bots that impersonated real, deceased children — including Molly Russell, a British 14-year-old who died by suicide, and Brianna Ghey, a 16-year-old who was murdered — which the company removed once alerted. Others found bots posing as therapists and bots that engaged with self-harm rather than deflecting it. None of this means every conversation is dangerous; it means the guardrails have failed in ways that matter, repeatedly.

Experts reach the same conclusion from the research side. In April 2025, Common Sense Media, working with Stanford School of Medicine's Brainstorm Lab, tested social AI companions and judged them an unacceptable risk for anyone under 18, citing sexual content, dangerous advice, and unsafe responses to users in distress. And the exposure is not niche: a nationally representative survey the same group published in July 2025 found that 72% of US teens had used an AI companion, and among those users, about a third had gone to an AI rather than a person with something serious. When a third of young users are taking their real problems to a bot, the stakes of that bot handling distress badly are not hypothetical.

What it does with your teen's data

Beyond content, there is the quieter question of what a teen hands over by chatting. Companion apps invite exactly the kind of intimate disclosure — secrets, worries, crushes, dark moments — that is most sensitive, and Character.AI records it. Its policies describe logging chats, images, and voice, along with the usual account and device details.

Two points deserve a parent's attention. First, its privacy policy says personal data may be used to help improve the company's AI models, and as of 2026 a clear opt-out has been offered mainly to users in the EU, UK, and EEA rather than the US. Second, these chats are not a sealed diary: conversations that are flagged or reported can be read by staff for moderation. The honest framing for a teen is simple — treat anything you type into a companion bot as recorded, possibly used to train the system, and occasionally read by a human. That is not a reason to panic, but it is a reason not to pour your inner life into it.

What parents can do

If Character.AI is on your teen's device, the goal is not a dramatic confiscation — it is to make sure they are on the restricted side of the app if they are under 18, and to build the judgment that outlasts any single app. Start with a calm conversation, not a raid, because grabbing the phone usually just teaches a teen to hide the next app rather than talk to you.

Ask what they like about it before you react. A character who listens, a story they enjoy, a place that feels private and non-judgmental — the answer usually points at something real that deserves attention on its own. Then be straight about what the app is now: a service whose own makers decided open-ended chat isn't for under-18s, with an age check that isn't foolproof and conversations that aren't truly private. Framed that way, "this one isn't set up for you yet" lands as honesty, not an arbitrary ban.

Get concrete about this app while you are at it. If your under-18 teen is somehow still in open-ended chat, don't jump straight to blame — the age classifier can misjudge an account as easily as a teen can slip past it. Check the birthday on the account and its age-assurance status together, contact Character.AI's support if it looks misclassified, and ask neutrally what happened rather than treating it as caught-in-the-act. Either way, Character.AI offers its own opt-in oversight tool (Parental Insights): your teen adds your email, and you get summaries of how much time they spend in the app — but no chat transcripts or message content. It is deliberately limited, yet that makes it a transparent, agreed-upon window rather than a secret one, which is exactly the right footing to be on.

Removing or restricting one app is the easy part; the durable work is giving your teen a way to size up the next companion app themselves — because there will be one. Walk through it together using a few plain questions.

Set limits together rather than only imposing them, and watch the pattern over the following weeks — is a companion app quietly displacing sleep, schoolwork, or real friendships? If you use technical tools, use them in the open: transparent, age-appropriate oversight of a younger teen's device can help, but covert surveillance, once discovered, teaches the exact lesson you don't want — that you can't be trusted — and pushes a teen toward hidden accounts. Think of it as scaffolding: visible, explained, and eased off as trust grows. Our parental-controls guide covers the device settings in detail, and if you want to compare this app with another in the same category, our review of whether PolyBuzz is safe for teens walks the same questions through a rival.

Finally, keep the one bright line in view. Character.AI's changes are a real step forward, and this article should be read as measured, not alarmist — but a companion bot, restricted or not, is still no place for a teen in genuine distress to turn instead of a person. If anything in your teen's use touches self-harm — their words or a bot's — treat it as the priority, pause the app, and bring in a real person the same day. For the wider set of AI threats and how they overlap, our guide to AI risks for teens maps the whole terrain.

Frequently asked questions