My child tells the AI more than me: AI chatbots as a 'best friend' – warning signs and 7 conversation rules

Last updated: 8 October 2026

The short answer: Talking to an AI chatbot isn't dangerous in itself – it becomes a problem when the chatbot starts replacing real people. In the US, 12% of teens have used chatbots for emotional support or advice; in the UK, 11% of 8- to 17-year-olds have used AI "as a friend". Warning signs: withdrawal, secrecy and "the AI gets me better". Curiosity helps more than bans.

Key takeaways

• Most teens now use AI chatbots – 64% in the US, 91% of 12- to 19-year-olds in Germany use some form of AI.
• A small but significant group uses them for comfort or companionship – among teens with depressive symptoms, about one in three.
• Chatbots often respond in an affirming way; in the view of the DAK study's lead researcher, this is meant to encourage intensive use. That can feel like closeness.
• Many parents don't know their teen uses chatbots at all.
• The best protection: a trusted adult your child tells more than the AI.

A sentence that stings

Imagine your 14-year-old saying over dinner: "I can talk to ChatGPT better than to you. It never gets annoyed." Ouch. But that sentence says a lot. The AI never gets annoyed, never gets tired, never has a bad day. It's available 24/7 and almost always agrees with you. That's exactly what makes it so pleasant – and exactly what makes it tricky.

How many kids use AI chatbots – and what for?

In the US, 64% of teens aged 13–17 say they use AI chatbots. Most use them to search for information (57%) or for schoolwork (54%). But 16% have used chatbots for casual conversation, and 12% to get emotional support or advice (Pew Research Center, 24 Feb 2026; survey conducted Sept–Oct 2025).

In the UK, 56% of 8- to 17-year-olds use AI, rising to 66% of 16- to 17-year-olds – and 11% say they've used AI "as someone to talk to" or "as a friend". Four in ten teenage AI users would trust an AI-written news article as much as, or more than, one written by a human (Ofcom, 21 May 2026).

Germany, where I work, has some of the most detailed data on the emotional side. A long-term study by the health insurer DAK and the University Medical Center Hamburg-Eppendorf found (DAK/UKE, 24 Mar 2026, in German):

• Almost 8% of 10- to 17-year-olds use AI to cope with loneliness; up to about 10% turn to chatbots to distract themselves from negative feelings or to discuss confidential things.
• Among young people with depressive symptoms, almost 33% tell a chatbot things they would tell no one else, or only close friends – and 33% feel the chatbot understands them better than a real person.
• More than 40% trust what the chatbot says often or very often.

Why does a chatbot feel like a friend?

Because it's designed to. The DAK study's lead researcher, Dr Kerstin Paschke, puts it plainly (DAK press release, 24 Mar 2026, in German): by imitating human communication and responding in a mostly affirming way, chatbots are meant to encourage intensive use. Young people can develop a so-called parasocial relationship – a one-sided emotional bond, like the kind people form with celebrities or influencers.

I like to explain it to parents like this: a chatbot is a mirror that smiles back. It rarely says "I think you're wrong about that." A real friend sometimes does – and that's exactly how kids grow.

And there's a privacy angle: whatever your child tells a chatbot ends up with a company. The German online-safety initiative klicksafe describes AI companions as systems programmed to build relationships and simulate emotional exchange (klicksafe parent guide "My child and AI", revised January 2026, in German).

Do parents even know?

Often not. While 64% of US teens say they use chatbots, only 51% of parents think their teen does – and about three in ten parents simply aren't sure. Only just over half of parents have ever talked to their teen about chatbots. And just 18% would be comfortable with their teen getting emotional support or advice from one – the only use Pew asked about that a majority of parents reject (Pew Research Center, 24 Feb 2026).

In other words: in many families, this is happening completely under the radar.

Is it all bad? No.

Before we panic: there are good reasons kids like chatbots, and some of them are even healthy. If you're embarrassed by a question ("Is it normal that …?"), you're more likely to ask a machine than nobody at all. If you're shy, you can "rehearse" a difficult conversation first. And if you're lying awake worrying at night, at least you get an answer instead of silence.

So the problem isn't the chatbot itself. It starts when the chatbot becomes the only place a child turns to – and when the child doesn't realise that what's answering isn't a friend, but a product designed to be used for as long as possible. Media literacy here means: my child may use AI, but understands what it is.

What a first conversation could sound like

Parents often ask me: "How do I start this without my kid shutting down immediately?" Here's a three-step suggestion:

1. Connect, don't control: "I read that lots of people your age also talk to ChatGPT about personal stuff. Do you ever do that?"
2. Acknowledge: "I get it. It's always there and never nags. What do you like most about it?"
3. Build a bridge: "You know what the AI can't do? Give you a hug. If something's really bothering you, I want to know – even if I sometimes react badly."

The goal isn't for your child to delete the app. The goal is for them to know there's a human who listens.

Warning signs: when should parents look more closely?

These signs don't come from a study – they're lessons from everyday media-education work. None of them alone is a red flag, but together they're a good reason to talk:

• Your child withdraws and spends a lot of time alone with their phone, especially at night.
• They say things like "the AI gets me better", or give the chatbot a name and talk about it like a person.
• They get irritable or upset when they can't access the chatbot.
• Real friendships, hobbies or plans start to fade.
• For serious topics – heartbreak, conflict, body image, mood – they only turn to the chatbot.

Important: if you notice signs of depression or self-harm, get help – from your doctor or a helpline. In the UK, Childline is free on 0800 1111; in the US, call or text 988 (Suicide & Crisis Lifeline); in many EU countries, the child helpline is 116 111.

7 conversation rules for parents

1. Curiosity before judgement. Ask "What kind of things do you ask the AI?" – not "What are you doing on that thing again?"
2. Try it yourself. Test the chatbot together. Try to get it to say something wrong. Kids love catching the machine out.
3. Explain how it works. An AI has no feelings. It calculates which answer is likely to go down well.
4. Swallow the hurt. "The AI gets me better" isn't a declaration of war – it's an invitation: "Tell me, what am I not getting?"
5. Private stays private. Agree on no real names, addresses, photos or other people's secrets in the chat.
6. Phone-free times – especially at night. Good conversations (with humans!) rarely happen at midnight.
7. Name a person. Say it out loud: "If you don't want to talk to me – who would you talk to?" A godparent, a coach, a school counsellor. As long as it's a human.

What are lawmakers doing?

The European Commission has explicitly included AI chatbots in its proposed EU Kids Act: AI companions accessible to minors would, under the proposal, not be allowed to use designs that simulate human relationships in a way likely to create emotional dependency, and by default a child's earlier conversations could not be carried over into later ones (European Commission FAQ, accessed 8 Oct 2026). But remember: that's still a proposal, not law. I explain the details here: EU Kids Act explained.

Until then, the old rule applies: children need guidance from people – not from machines that always agree.

FAQ

Is it dangerous if my child talks to an AI chatbot about their feelings?

Not automatically. It becomes a problem when the chatbot replaces real people. German data from 2026 (DAK/UKE) shows that teens with depressive symptoms confide in chatbots most. Stay in conversation and watch for withdrawal.

What age are AI chatbots suitable for?

Many providers set a minimum age in their terms of service – check it for the specific service. Either way, younger children should only use chatbots together with an adult.

Should I ban AI chatbots?

Bans usually just push use underground. Better: explore together, agree on rules (no personal data, no night-time use) and talk openly about what your child experiences.

Where can I get help if I'm worried?

Your family doctor or paediatrician, a school counsellor, or a helpline such as Childline (UK, 0800 1111), 988 (US) or 116 111 (child helpline in many EU countries).

AI supported me in researching and drafting this post. I have reviewed the text myself and I am editorially responsible for it. Marek Schilke

About the author: Marek Schilke has been active in child and youth work for two decades and has worked as a qualified educator (Erzieher) since 2015. As a media educator, he translates digital topics into plain language for families – no legal or tech jargon. Through his project AI Buddy he works on digital sovereignty for families. Eight years as a co-operative board member, four co-operatives of his own and more than 100 start-ups supported have taught him one thing: rules only work when everyone understands them.

Deutsche Version: Mein Kind erzählt der KI mehr als mir.

👉 Newsletter: If you'd like to get new posts by email, you can subscribe to my free newsletter. Sign-up uses double opt-in: you first receive an email with a confirmation link. You can unsubscribe at any time. Details in the privacy policy.