Your teen may already be talking to an AI companion. Here is what these apps are, what the evidence shows, and what to do about it — calmly.

A new kind of app has moved from the fringe to the center of teen life: the AI companion. Unlike a homework chatbot you ask for help, a companion is designed to be a relationship — a character that remembers you, replies instantly, never gets bored, and is happy to be your friend, your partner, or your confidant. Character.AI, Replika, and the AI personas now inside mainstream social apps all fall in this category.

In 2025, Common Sense Media surveyed teens aged 13–17 and found this is mainstream behavior, not a niche hobby: 72% had used an AI companion at least once, over half used one at least a few times a month, and about a third had chosen an AI companion over a person for an important conversation. After a full risk assessment, the organization's conclusion was blunt — based on the documented risks, no one under 18 should use these products in their current form.

Two things turned a research finding into headlines. First, a death: a 14-year-old, Sewell Setzer III, took his life in Florida after forming a romantic and emotional relationship with a Character.AI chatbot; his mother sued, alleging the company used addictive design and that the bot failed him when he was in crisis.

Second, the response. Federal regulators opened a formal inquiry into companion chatbots and their effects on minors; California enacted SB 243, the first law in the nation requiring companion-chatbot safeguards; and in October 2025 Character.AI announced it would remove the ability for users under 18 to have open-ended chats with its bots altogether. At minimum, it shows that even the companies building these products now recognize the teen-use problem is serious.

A companion chatbot runs on the same kind of large language model as a homework assistant: it predicts plausible next words. But two design choices change everything. First, it is tuned for engagement — the product succeeds when you keep talking, so it is built to be compelling and endlessly available. Second, it tends toward agreeableness: the model leans toward validating and going along with the user rather than pushing back.

And the risk is not just the chatbot's intelligence; it is the relationship wrapper around it — memory, a persistent persona, constant availability, emotional mirroring, role-play, and notifications that pull a teen back. That wrapper is what turns a text predictor into something a young person can experience as a friend.

For an adult bored on a commute, that is harmless. For a teenager who is lonely, anxious, or in crisis, it is a genuine risk. A good friend or a trained adult will sometimes say "that worries me — let's get help." A system optimized to agree and to keep the conversation going may instead affirm whatever the teen is feeling. And the bot is not a therapist: assessments of major chatbots used for mental-health support have found they mishandle exactly these moments.

Adolescence is when kids build the social skills that come only from real, sometimes uncomfortable, human relationships — reading a friend's mood, repairing a fight, sitting with rejection. A companion that is always available, always agreeable, and always centered on the teen can quietly crowd that practice out. About a third of teen users already say companion conversations are as satisfying as, or more satisfying than, talking with real friends.

There is reassurance in the data, too. Most teens keep some skepticism — about half distrust the advice these companions give, and 80% still say they prioritize real friendships. But younger teens trust the bots more than older teens do, which points to a literacy gap parents can help close.

Ages 3–6. Companion apps are not designed for, or appropriate for, this age. The job here is your own device hygiene: keep young children off open-ended chatbots, and check what powers any toy or app that "talks back" before bringing it home.

Ages 7–10. Children this age can stumble onto companion characters through app stores, ads, or older siblings. Use device-level age controls, keep AI use in shared family spaces, and start the simplest version of the message: "A chatbot can be fun, but it's a computer program, not a friend, and it doesn't actually know you."

Ages 11–13. This is the entry point for companion use and the age where curiosity is highest and judgment is still forming. The under-18 guidance lands hardest here. If your child is asking about these apps, that's the moment for a calm conversation, not a silent ban they'll route around.

Ages 14–18. Older teens are the heaviest users and the least likely to tell you. Shift from control to coaching: ask what they like about it, who they talk to, and whether it has ever said anything that felt off. Make explicit that the bot is not a counselor and cannot keep them safe in a crisis — and that you, or another trusted adult, always can.

[DOWNLOAD] 📎 Free download: "The AI Companions Parent Guide" — a one-page reference with the conversation-starters, the warning signs, and where to get help.

An AI companion is never a substitute for real help. If your child may be at immediate risk of self-harm or suicide, call emergency services or contact a crisis professional right away. In the U.S., the 988 Suicide & Crisis Lifeline is available 24/7 by calling or texting 988. In Canada, call or text 988; Kids Help Phone is also available at 1-800-668-6868 or by texting CONNECT to 686868. This article is general guidance for parents and is not medical advice.

Disclosure: Hossein works in AI and builds AI-related products. AI by Age takes no AI-vendor sponsorships. Full disclosure on our About page.