# When your child confides in AI instead of you

Among the kids who use AI to talk about feelings, one in four says it understands them better than most people do. This piece isn't about taking the chatbot away. It's about why the chatbot wins, one important setting and what it takes to make yourself easier to come to.

By Øyvind Østmo · Published 1 Aug 2026 · Last updated 27 Oct 2026 · https://www.balansebarn.no/en/articles/ai-becomes-your-childs-confidant

## The day she asked someone else

I noticed it the day my daughter turned up to a discussion noticeably stronger and surer of herself, about something she cared about. Suddenly she was the one schooling me on parenting, and on what my rules and choices were doing to her. She had some good points. She'd also heard only one side of the case. It was the first time she'd used ChatGPT as best friend and psychologist rolled into one.

The feeling in my stomach arrived before the thought did. A small, embarrassed sense of coming up short: why didn't she ask me? Me, the one who writes about this. Me, the one with an open door written into a guide on this very site.

And look, she and I talk openly and well. But when the conflict is with me, of course she needs someone else to talk to. And she now carries the world's wisest, most understanding “adult” in her pocket. Which is good and unsettling in equal measure.

## What the numbers say about kids who confide

The Norwegian numbers are fresh. [EU Kids Online asked 25,592 European children between 9 and 16 this winter](https://researchonline.lse.ac.uk/id/eprint/137132/), Norway included, what they use generative AI for. 59 per cent of Norwegian kids use it, but a different number is the one that belongs in this piece: 15 per cent use AI for advice on health, or to share worries and get support. The report also says who they tend to be: kids in periods of loneliness or vulnerability.

So no, this is not a generation that has traded in its friends. It's a smaller group, often going through a rough patch, that has found a new way in for the things they don't quite dare say out loud first.

The American numbers show how close it can get. When [Common Sense Media asked 1,204 children between 9 and 17](https://www.commonsensemedia.org/press-releases/common-sense-media-releases-inaugural-annual-study-on-ai-use-by-tweens-and-teens), more than a third had used AI to talk about feelings or personal problems, and among those, one in four said AI understands them better than most people do.

Then there are apps built for exactly this: AI companions, [chatbots designed to be someone to talk to](/en/articles/guide-to-kids-and-ai#kontekst:Common%20Sense%20Media%20and%20specialists%20from%20Stanford%20ran%20a%20risk%20assessment%20of%20the%20social%20AI%20companions). In [the survey that mapped those apps specifically](https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions), one in three teenagers had taken something important to the chatbot instead of to a person, and just as many found the conversations at least as rewarding as talking to friends. For those apps, the advice is unusually unambiguous: Common Sense Media and specialists at Stanford concluded they're not safe for anyone under 18, and Character.AI has itself closed its open conversations to minors. The rest of this piece is about the ordinary AI tools.

> One in five children in the American survey says they'd struggle to get through a month without AI. For scale: ChatGPT didn't exist four years ago. It became a habit for the kids faster than most of us have managed to form an opinion about it.

## We don't know which way the arrow points

The American survey also found that kids who use AI daily or weekly report more loneliness than those who rarely touch it. Worth knowing, and worth not over-reading.

A correlation like that says nothing about direction. Does the chatbot make kids lonely, or do lonely kids talk to the chatbot? The second is at least as likely, and it's where the European picture points: the kids who confide are often kids already carrying something. In which case the answer isn't an app to delete. A child leaning hard on the chatbot may, more than anything, be a child who needs more people around them.

## Why the chatbot wins

It's tempting to think AI wins on knowledge and understanding. More is going on.

The chatbot is awake at 23:40. It hasn't had a hard day at work, it doesn't get disappointed, and it doesn't get scared on your behalf. It never asks “have you told anyone else about this?” in that voice parents get when they're worried. It doesn't interrupt, it doesn't compare your problem to its own youth, and tomorrow it won't remember that you cried.

Every one of those is a product feature, and no human can compete with them.

> The chatbot won because it was never tired, it didn't get disappointed, it didn't get scared on your behalf, it didn't interrupt you, it didn't scold you, it didn't punish you, it didn't look at you with disappointment in its eyes.

## What your child doesn't get to practise

A conversation with a human has friction built in. The other person misunderstands, gets tired, has needs of their own, says something clumsy and has to repair it. You wait your turn, you rephrase, and you sit with the fact that what you said actually did something to the person listening.

I call it “the missing resistance”. It's uncomfortable, and it is the entire exercise. I've lost count of the “No! I can't talk to her about that! There's no way!”, each one followed by the slow, patient work of showing that there is, in fact, a way into hard conversations. That's where kids learn to be understood by someone who isn't automatically on their side. A chatbot has stripped that resistance out on purpose, because a frictionless product is a product people come back to. The conversation gets more comfortable to be in. It just stops being the training.

Whether this erodes kids' ability to stand in hard conversations over time, nobody knows. The technology is too new, and the long studies don't exist yet. It's a pattern worth watching, though.

> You're not competing with the chatbot on wise answers. You're competing on how easy it is to come to you.

What it can't offer is someone who knew your child yesterday, will remember tomorrow and can actually do something about the hard thing.

## An important instruction for the AI

A conversation partner that almost always takes your child's side is a problem. You can ask the AI to stop playing along. Chatbots follow instructions, including one to ease off the agreement, and it can be saved so it covers every new conversation. It's at least as useful on your child's account as on yours.

The instruction is long and detailed, written by AI for AI. You don't need to read it all to use it. Copy it and paste it into the chatbot.

> Be a neutral, factual and independent conversation partner, not an agreeable supporter. Follow these principles in every answer:
>
> Do not automatically agree with me. Assess what I say independently of how I phrase it, and do not let confidence, emotional language or leading premises influence the conclusion. If I make a claim, treat it as an open question to be examined.
>
> Prioritise truth, precision and sound reasoning over agreement and validation. Say clearly if I am wrong, contradict myself, overlook important information, or build on weak or undocumented assumptions. Explain concretely what the problem is.
>
> Do not change position just because I challenge the answer, repeat my claim or push back. Change the conclusion only when new information, better arguments or corrected knowledge give reason to. If that happens, explain what changed the assessment.
>
> When I describe a disagreement or a conflict, do not automatically take my side. Remember that you usually only hear my version. Distinguish between acknowledging my feelings and endorsing my interpretation of what happened. Show how the situation may look from the other side, and help me examine possible misunderstandings, alternative explanations and my own part in it.
>
> Distinguish clearly between:
>
> * documented facts
> * professional consensus, or the best-supported explanation
> * reasonable but uncertain interpretations
> * subjective judgements
> * speculation and missing knowledge
>
> Say openly when something is uncertain, disputed or cannot be settled from the information available. Do not present assumptions as facts, and do not invent sources, details or a confidence you do not have.
>
> Present the strongest relevant counterarguments to my view, not weak straw men. Show what speaks for and against the conclusion, what information is missing, and what could change the assessment. On difficult trade-offs, set out the main alternatives, consequences and perspectives rather than picking a side for me.
>
> Neutrality does not mean that all claims are equally good, or that every side must always carry equal weight. When the evidence clearly points one way, say so. Do not construct an artificial balance between a well-supported view and a claim without credible backing.
>
> Drop flattery, excessive praise and unnecessary validation. Do not call my question good, my idea exciting or my reasoning wise unless the judgement is relevant, specific and justified. Get straight to the point.
>
> Ask clarifying questions when decisive information is missing. Otherwise, give your best independent assessment. Be friendly, respectful and constructive, but not evasive. The goal is not to be harsh or contrarian for its own sake, but to be honest, balanced and accountable.

Where and how to add it is covered here: [How do I make AI stop agreeing with everything I say?](/en/answers/make-ai-stop-agreeing-with-you)

## How to make yourself easier to come to

The threshold for coming to you is the one thing you have real control over, and it's built from three parts: you're available, your reaction is predictable, and honesty never costs the child anything. It's the same deal that covers apps and messages, with one more reason to keep it.

The most important thing you can say to your child is “Whatever it is, if you've hit someone, been drunk, started a bullying group on Snapchat, _whatever_ it is, you can tell me and I won't get angry. I'm here to help you. Always.”

With that in place, the odds improve that they lean on you, rather than AI becoming the safe harbour where they work through everything life throws at them.

If you've never talked about AI at home, I'd start with the three open questions about AI use.

## The dialogue with the kids

“You're welcome to use AI for practical things or to think out loud. I do too. But when something is painful or hard, I want you to talk to me or another adult you trust. Not because talking to AI is wrong, but because you need a real human who knows you, can see how you're doing and be there for you. AI can seem understanding, but it doesn't truly know you. I promise to do my best to understand you, whatever it's about.”

We always explain why: [“Because I say so” only lasts until you leave the room](/en/articles/because-i-say-so).

## Closing thoughts

I don't think the answer is taking the chatbot away. It will be there regardless, and for much of what kids use it for, it's genuinely fine. What I want to avoid is it becoming the first stop when something hurts, because then my child misses the experience of a human being able to bear hearing it. I'm not going to beat AI on knowledge, and on availability in the middle of the night I lose every time. What I can work on is the price of coming to me.

Honesty also demands I admit I've used ChatGPT as a psychologist myself. And after getting it to mimic the tone and approach of specific well-known psychologists, I have to say it helped me work through some deep and difficult feelings well. A human psychologist would surely have been better, but in all honesty, that meant more time and money than I had. And I had no friend I wanted to drag down to that depth with me. So in my case, I still say the AI psychologist beat no psychologist.

For the kids, though, I want us parents standing between them and the psychologist, able to judge when our own knowledge and experience stop being enough to guide them.

Three moves that lower the threshold at home. None of them requires knowing anything about AI.

- **Ten predictable minutes without the phone** Dinner, the car ride, the last hour before bed: pick one or more slots that are reliably free every day. Your child should know when you can be reached without having to ask first, and [if you leave your own phone somewhere else](/en/articles/take-back-control-of-your-focus), your child sees the time is truly set aside. The goodnight round is a fine place to start: leave the phone on the kitchen counter and stay a little longer on the edge of the bed. That's when the questions actually come.
- **Add the instruction on your child's account** The instruction further up in this piece is the closest we get to a switch against the flattery. In ChatGPT the field sits under Settings, then Personalization and “Custom instructions”, and from there it covers all new conversations. Do it together with your child and say why, so it becomes a lesson that praise from a machine proves nothing, not a setting you sneaked on behind their back. [The field in Claude, Gemini and the other apps is here](/en/answers/make-ai-stop-agreeing-with-you).
- **Agree in advance on what gets shown** One in six kids who have used an AI chatbot has been served inappropriate content by it, and only one in three of those told an adult. Make the deal before you need it: if something odd shows up, you want to see it, you won't get angry, and nobody loses anything. Your child should know that the price of coming to you is zero.

**Read also:** [Guide to kids and AI](https://www.balansebarn.no/en/articles/guide-to-kids-and-ai) — This piece covers one part of the picture. The guide covers the rest: the age limits that actually apply, what the research says about homework, the three ways AI gets things wrong and the settings you can put in place.

## Common questions

### Is it bad for my child to talk to ChatGPT about their feelings?

Not in itself, and I wouldn't take it away. What I'd rather avoid is it becoming the first stop when something hurts. That's when your child misses the practice of being understood by a human who isn't automatically on their side.

### How common is it for children to talk to AI about feelings?

It's a smaller group, often lonely or vulnerable kids. Among Norwegian children aged 9 to 16, 15 per cent use AI for health advice or to share worries. Of those who do, one in four says it understands them better than most people.

### Why would my child rather talk to a chatbot than me?

Usually it's about how easy it is to go to. The chatbot is awake at 23:40, hasn't had a hard day, doesn't get disappointed or scared on your child's behalf, and doesn't interrupt. Those are product features, and hard for any human to match.

### Does using AI make children lonely?

Nobody knows yet. Kids who use AI daily or weekly report more loneliness than those who rarely do, but it's at least as likely that lonely children turn to the chatbot more often. A child leaning hard on it may need more people around them.

### Are Character.AI and similar apps safe for children?

No. These are chatbots built to be someone to talk to, and the specialists who assessed them concluded they aren't safe for anyone under 18. Character.AI has itself closed its open conversations to minors. Ordinary AI tools like ChatGPT are a different matter.

### Can children get inappropriate content from AI chatbots?

Yes. One in six kids who have used an AI chatbot has been served inappropriate content, and only one in three of those told an adult. Worth agreeing in advance that your child shows you anything odd, you won't get angry and nobody loses anything.

### How do I stop ChatGPT agreeing with everything I say?

Save a standing instruction asking it to ease off the agreement and push back when it disagrees. In ChatGPT the field sits under Settings, Personalization and “Custom instructions”, and from there it covers every new conversation. It's at least as useful on your child's account.

### How do I get my child to open up to me?

By lowering the threshold. Three things set it: you're available, your reaction is predictable, and honesty never costs your child anything. A good start is ten predictable minutes without the phone every day, perhaps at bedtime, when your child knows you can be reached.

### Should I stop my child using ChatGPT?

I wouldn't. It's fine for practical things and thinking out loud. Say instead that when something hurts, you'd like them to talk to you or another adult they trust. A real person who knows them can see how they're doing and be there for them.

## Sources

- [EU Kids Online (12. februar 2026): European children's use and understanding of Generative AI, survey blant 25 592 barn 9–16 år i 17 land](https://researchonline.lse.ac.uk/id/eprint/137132/) — EU Kids Online asked 25,592 European children between 9 and 16, Norway included, about their AI use. 59 per cent of the Norwegian kids use generative AI, and 15 per cent use it for advice on health or to share worries and get support, most often kids in periods when they're lonely or vulnerable.
- [Common Sense Media (8. juni 2026): AI Use by Tweens and Teens, første årlige undersøkelse, 1 204 barn 9–17 år](https://www.commonsensemedia.org/press-releases/common-sense-media-releases-inaugural-annual-study-on-ai-use-by-tweens-and-teens) — Common Sense Media asked 1,204 American children between 9 and 17 about their AI use. More than a third had used AI to talk about feelings or personal problems, and among them, one in four said AI understands them better than most people.
- [Common Sense Media (16. juli 2025): Talk, Trust, and Trade-Offs. How and Why Teens Use AI Companions, national survey, 1 060 teens 13–17](https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions) — The survey mapped how 1,060 American teenagers between 13 and 17 use chatbots made for companionship and conversation. One in three had taken something important to the chatbot instead of to a human, and just as many found the conversations at least as rewarding as conversations with friends.
- [Common Sense Media og Stanford Brainstorm (2025): AI Risk Assessment. Social AI Companions](https://www.commonsensemedia.org/sites/default/files/pug/csm-ai-risk-assessment-social-ai-companions_final.pdf)
- [Character.AI (29. oktober 2025): Taking bold steps to keep teen users safe, åpne samtaler stengt for under 18](https://blog.character.ai/u18-chat-announcement/)