Ask most parents what their child does online and they will mention games, videos and messaging. Fewer would say "talks to an AI". Yet AI chatbots have quietly become one of the most common things children do online, and for some, the bot has started to feel less like a tool and more like a friend. This is a calm look at what is happening and what, if anything, you should do about it.
What the research shows
Internet Matters, the UK online safety charity, surveyed 1,000 children and 2,000 parents for its report on children and AI chatbots. The findings are striking.
- Nearly two thirds of children (64 per cent) now use AI chatbots.
- More than a third of those who use them (35 per cent) say talking to a chatbot feels like talking to a friend.
- One in eight children (12 per cent) use a chatbot because they feel they have no one else to talk to.
The picture is more worrying for children the charity describes as vulnerable. Among that group, 71 per cent use chatbots, half say it feels like talking to a friend, and roughly a quarter say they would rather talk to a chatbot than a real person.
Separate research shared by Vodafone Three around Safer Internet Day 2026 pointed the same way, finding that a large share of children who use AI chatbots think of them as friends. This is not a fringe habit. It is becoming normal.
Why children lean on them
It helps to understand the appeal before deciding how to respond. A chatbot is available at three in the morning. It never gets bored, never judges, and never tells anyone. For a child who is anxious, lonely or simply working something out, that can feel safer than talking to a person.
Chatbots are also genuinely useful. In the same research, more than four in ten children said they used them for schoolwork, and most felt asking a chatbot was easier than searching for themselves. The problem is not that children use AI. It is when a chatbot starts to fill a gap that a trusted person should be filling.
Where the risks lie
There are a few things worth having in mind.
- It sounds confident, even when it is wrong. A chatbot can give inaccurate or inappropriate answers in the same warm, certain tone it uses for everything. Internet Matters found many child users had no concerns at all about following a chatbot’s advice.
- It can crowd out real support. If a child turns to AI during a hard moment instead of a parent, teacher or friend, they may miss the human help they actually need.
- It is not a person, however human it feels. A chatbot has no genuine understanding of your child, and it is built to keep the conversation going, not to know when to say "you should talk to an adult about this".
There is also an awareness gap at home. In the Internet Matters survey, six in ten parents worried their child thinks chatbots are real people, yet only about a third had actually talked to their child about how to judge whether AI information is true.
What to do about it
You do not need to ban chatbots, and for most families that would not work anyway. The aim is to keep AI in its place: a helpful tool, not a substitute friend or a source of life advice.
- Get curious, not cross. Ask which chatbots your child uses and what for. A genuine "show me how it works" opens the door far better than a warning.
- Name the difference out loud. Explain, simply, that a chatbot predicts likely words rather than understanding them, and that it can be confidently wrong. Children handle this idea well once someone says it.
- Agree what it is good for. Homework prompts, ideas and explanations, yes. Decisions about friendships, health, feelings or anything that matters, that is a job for a real person.
- Keep the human door open. The strongest protection is a child who knows they can come to you. If a chatbot is filling a loneliness gap, that is the thing worth gently exploring, not the app itself.
- Check the age rating. Most mainstream chatbots are intended for teenagers, not young children, and some companion style apps are for adults only. It is worth knowing what your child is actually using.
The bigger point
Children treating a chatbot as a friend is not a sign that something has gone wrong with your child. It is a sign of how convincing this technology has become. The healthiest response is not fear but honesty: use these tools with your child, talk about what they can and cannot do, and make sure the real people in their life are always the easier ones to turn to.