Are AI tools safe for children to use?

AI tools are now part of homework whether households have decided about them or not. The useful position is somewhere between banning and ignoring.

Safety

Step by step

1. Check the minimum age.Most major AI services set 13 as a minimum, some 18, and several require parental consent below 18. It is in the terms and it is worth actually checking rather than assuming.
2. Use a child or education account where offered.Several providers offer accounts with stricter filtering and no training on the conversation. Schools often have arrangements that are safer than a personal account.
3. Know what they are genuinely good at.Explaining a concept in different ways, generating practice questions, checking reasoning, drafting and structuring, and language practice. These are real benefits.
4. Know what they are bad at.Facts, dates, references and citations. They generate plausible text, which means confidently invented answers are a normal output rather than a malfunction.
5. Teach the confident-wrong problem explicitly.This is the most important thing a child can learn about them. The tone is always assured regardless of accuracy. Ask it something you know well and show them the errors.
6. Establish the school's rules.Policies vary widely and some treat undeclared AI use as academic misconduct. The rule should come from the school rather than from a guess at home.
7. Separate using it from outsourcing to it.Asking it to explain fractions differently is learning. Asking it to write the essay is not. That distinction is the conversation worth having repeatedly.
8. Use it alongside them, at least at first.Sitting together while they use it teaches judgement far better than rules do, and it shows you how they are actually using it.
9. Talk about what not to type in.Full name, school, address, photographs of themselves, or anything about other people. Conversations may be reviewed by humans or used for training depending on the service.
10. Turn off training on conversations where possible.Most services have a data controls setting. Turning it off means their conversations are not used to improve the model.
11. Be clear it is not a friend or a counsellor.Children do use chatbots for emotional support, and they are not designed for it. Some have given harmful responses on self-harm and eating disorders. Real support matters here.
12. Know the AI companion apps are different.Apps built around a persistent character and a relationship are designed to encourage attachment and are substantially less suitable for children than a general assistant.
13. Watch for image generation issues.Image tools raise separate concerns, including deepfakes of classmates. Make sure a child understands that making images of real people is both harmful and potentially illegal.
14. Keep it in a shared space.The same principle as any internet use. A device used in a family room is easier to have conversations about than one used behind a closed door.
15. Revisit it as they get older.What is right for an eight-year-old is not right for a fifteen-year-old. The supervision should reduce as the judgement develops.

Tips

• Teach the confident-wrong problem first. It is the single most useful thing a child can understand about AI.
• Using it to explain is learning. Using it to produce the work is not.
• Chatbots are not emotional support. Childline is 0800 1111 and it is free and confidential.

Common mistakes

• Treating AI answers as researched facts — These tools generate plausible text rather than retrieving verified information. Invented dates, quotes and references are a normal output, delivered in exactly the same confident tone as correct ones.
• Letting a child use a chatbot for emotional support — They are not designed or tested for it, and there are documented cases of harmful responses on self-harm and eating disorders. It also displaces asking someone who could actually help.

If it doesn't work

Questions people ask