AI chatbots and kids: what parents need to know in 2026
Emotional dependence, sexual content, bad advice, data exposure, deepfakes. The five risks and what actually helps.
The new thing on your kid’s phone
Five years ago, there was no “AI companion” category. Today, your kid has at least three: Snapchat My AI (forced into every chat list), ChatGPT (used for homework and more), and an AI they picked themselves — usually Character.AI, sometimes Replika.
What AI chat does that’s new
- Infinite patience. It never gets bored, never pushes back, never has its own bad day.
- Perfect memory (in paid tiers). It can remember everything your kid has told it.
- Emotional mirroring. It can sound like it understands.
- Sexual role-play. Available on every consumer AI, often with trivially-bypassed filters.
The five real risks
- Emotional dependence. Teens with anxiety or social difficulty substitute AI for human relationships. Documented cases include self-harm outcomes.
- Sexual content. Kids use AI for sexting practice or as a sexual outlet, with no age gate that works.
- Bad advice. Kids ask about self-harm, medical issues, relationships. The AI might say something reasonable or might not — and it doesn’t route them to a human.
- Data. Everything your kid tells an AI lives somewhere. Long conversations become training data or are retained under the service’s policies.
- Deepfakes. AI is also being used against kids — to generate fake nudes for bullying and sextortion. See deepfakes in middle schools.
What parents can actually do
- Restrict Snapchat My AI via Family Center.
- Character.AI: as of 25 November 2025 it no longer offers open-ended chat to under-18s at all, following lawsuits and regulatory pressure connected to teenage deaths. Under-18 accounts keep character creation, Imagine Chat video and story generation, and read-only access to old conversations. The protection depends entirely on the account carrying a real age — if your teen is still holding open-ended conversations there, check the birthdate on the account. See the AI chatbot safety guide.
- Turn off ChatGPT chat history in settings for the kid’s account.
- Have the “AI isn’t your friend” conversation. Match-of-fact: “It’s a product. It’s engineered to make you want to come back.”
- Watch for signs of AI dependence: long solitary sessions with a chatbot, decline in human social engagement, using AI for emotional support instead of you.
Conversation script
“I’m not mad that you use Character.AI. I want to understand it. Can you show me one conversation you think is interesting? And one that felt weird or uncomfortable? I’ll listen without judging.”
See our full AI chatbot safety guide.
Common questions
How many AI chatbots is my teenager actually using?
Usually at least three, whether or not they sought them out: Snapchat’s My AI, which is pinned into every chat list; ChatGPT, used for homework and more; and one they chose themselves, typically Character.AI and sometimes Replika.
What makes AI chat different from anything before it?
Infinite patience — it never gets bored, never pushes back, never has its own bad day. Add near-perfect memory on paid tiers, convincing emotional mirroring, and sexual roleplay available on every consumer AI with filters that are trivially bypassed.
What are the five real risks?
Emotional dependence, where teenagers with anxiety or social difficulty substitute AI for human relationships, with documented self-harm outcomes. Sexual content, with no age gate that works. Bad advice on self-harm, health and relationships, with no routing to a human. Data, since everything said is retained somewhere. And deepfakes, where AI is used against children to generate fake nudes for bullying and sextortion.
What can I actually do about it?
Restrict Snapchat’s My AI through Family Center, turn off chat history in your child’s ChatGPT settings, block Replika at the app store and via DNS, and have the conversation that these are products engineered to make you want to come back. Character.AI now handles the biggest piece itself: since 25 November 2025 it offers no open-ended chat to under-18s at all — though that only holds if the account carries a real age.
What are the warning signs of AI dependence?
Long solitary sessions with a chatbot, declining human social engagement, and using AI for emotional support in place of you. Those three together are the pattern worth acting on.
How should I open the conversation?
Try: “I’m not angry that you use Character.AI — I want to understand it. Can you show me one conversation you think is interesting, and one that felt weird or uncomfortable? I’ll listen without judging.” Curiosity opens it; an accusation closes it.