Jump to a section
Most children already use AI chatbots, and many enjoy them. Check the app's age limit, switch on parental controls where they exist, and talk to your child about what AI is: helpful, often right, sometimes wrong, and not a person. Agree a few family rules, and keep the conversation going rather than banning it outright.
If you're worried your child is struggling, don't leave it to an app. Talk to them, and contact their GP or school. In an emergency, call 999.
Where children are with AI
If your child uses AI, they're far from alone. Ofcom found that 56% of 8 to 17-year-olds in the UK use AI, rising to 66% of 16 to 17-year-olds. A 2025 survey of 1,000 children aged 9 to 17 by the charity Internet Matters put it at 64%.
Ofcom also found that 52% of teenagers feel confident recognising AI-generated content, yet a quarter of those confident teenagers couldn't correctly identify an AI-generated image they were shown. Confidence and skill aren't the same thing, which is where you come in.
For most, it's homework, curiosity and fun. But Internet Matters also found that 35% of children who use chatbots say talking to one is like talking to a friend. Among vulnerable children, that rose to half, and 23% said they used chatbots because they had no one else to talk to.
None of this means AI is bad for children. It means it's worth knowing what your child uses, and talking about it.
Ofcom: the rise of AI in children's media lives (May 2026) Internet Matters: Me, Myself and AI (July 2025)
The age limits
Each app sets its own rules, and they differ more than you might expect. Here's where the main ones stand for families in the UK.
| App | The rule | Parental controls |
|---|---|---|
| ChatGPT | From age 13; under-18s need a parent or guardian's permission | Yes, when accounts are linked |
| Gemini | Google doesn't offer Gemini on supervised accounts for under-13s in the UK | Not available for under-13s in the UK |
| Claude | You must be at least 18 | Not applicable |
| Character.AI | Under-18s can no longer have open-ended chats with characters | The app uses age checks |
Setting up ChatGPT's parental controls
Link accounts in ChatGPT on the web; your teen must accept.
- Open Settings from your profile, then Parental controls.
- Select Add family member; invite your teen by email or phone.
- After they accept, select their name under Family members.
- Review the controls together. Either person can unlink; you are notified if your teen unlinks.
Which settings to look at
Four starting points from OpenAI's current help page:
- Reduce sensitive content: enabled by default.
- Improve the model for everyone: enabled by default; turn it off if you do not want those chats used to improve models.
- Voice Mode and Image generation: enabled by default; each can be disabled.
- Quiet hours: off by default; set times when ChatGPT cannot be used.
What you will and won't see
You cannot read chat history through parental controls. Safety notifications cover limited serious concerns and may miss problems; they are not live monitoring. Choose notification preferences in the linked-account settings.
Keep talking with your teen; the controls supplement that relationship.
When AI becomes a 'friend'
Chatbots are endlessly patient and always available, so it's no surprise some children treat them as a friend. Internet Matters found that 26% of vulnerable children who use chatbots would rather talk to one than a real person.
Some apps are built for exactly this kind of relationship. Character.AI, one of the best known, announced that under-18s would no longer be able to have open-ended chats with its characters from 25 November 2025, as Fortune reported.
You don't need to panic if your child enjoys chatting to an AI. Do keep an eye out for it replacing real friendships, or becoming the place they take worries. Google puts it well in its advice to parents: help your child understand that a chatbot is 'not a person and cannot think for itself or feel emotions'.
Internet Matters: Me, Myself and AI (July 2025) Fortune: Character.AI bans teens from open-ended chats Google: Gemini for children under 13
Teaching them that AI can be wrong
This is the most useful thing you can teach. Internet Matters found that 40% of children who use chatbots have no concerns about following their advice, and another 36% weren't sure. Only 34% of parents had talked to their child about how to judge whether what AI tells them is true.
Make it a game. Ask a chatbot something you both know the answer to, such as your town's football team or the date of a family birthday, and see whether it gets it right. Spotting a wrong answer together teaches more than any warning.
Can teachers tell if you used ChatGPT? Do AI detectors work?
Homework: make it a tutor, not a shortcut
The difference is in the question. "Write my essay on the causes of the First World War" skips the learning. "Quiz me on the causes of the First World War and tell me what I've missed" is learning.
Sit with your child the first few times and show them the difference.
- AI can explain, quiz and give feedback. It doesn't write the answer.
- We check what school allows before using AI for any piece of work.
- If AI helped, we say so when the school asks.
- We check facts from AI in a book or a trusted website.
- No names, photos or private details go into a chatbot.
I'm 13 and learning about photosynthesis for a test. Don't give me the answers. Ask me one question at a time, wait for my answer, then tell me what I got right, what I got wrong and give me a hint before the next question.Check what school allows
Schools set their own rules for homework, so ask your child's teachers what's allowed. For formal assessments the rules are stricter. JCQ, which sets assessment rules for UK exam boards, says teachers and assessors are key to ensuring only a student's own work is accepted for assessment.
If your child uses AI for anything that's marked, the safest habit is to be open about it and follow the school's guidance on how to say so.
Can teachers tell if you used ChatGPT? Do AI detectors work?
Starting the conversation
Curiosity works better than rules handed down. Try these over tea or on the school run:
- "Which AI apps do your friends use? What do they use them for?"
- "Has it ever told you something that turned out to be wrong?"
- "What's the best thing you've made with it?"
- "What would you do if a chatbot said something that upset you?"
Do it together
The best way to shape how your children use AI is to use it with them. Plan a day out, invent a story or get help with a tricky bit of maths, side by side. They'll see you check things, keep personal details out and use it as a tool, which is worth more than any setting.
Five family AI rules
- We use AI apps that are meant for our age, with parental controls on where they exist
- We never share our full name, address, school or photos with a chatbot
- AI can be wrong, so we check anything important with a person or a trusted website
- AI is a tool, not a friend. If something's worrying us, we talk to a real person
- Homework help is fine. Handing in AI's work as our own isn't
Sources
These sources support the guide. Tool settings, prices and policies change, so dated source boxes show when those details were checked.
- Internet Matters: new report reveals how risky and unchecked AI chatbots are the new go-to for millions of children (13 July 2025)
- Fortune: Character.AI bans teens from talking to its chatbots (29 October 2025)
- Google: Gemini Apps for your child under 13
- OpenAI Help: Managing parental controls in ChatGPT
- OpenAI: Europe Terms of Use (UK age and parental permission)
- Anthropic: Consumer Terms of Service
- Character.AI: Continuing to build upon our safety priorities (3 September 2026)
- JCQ: changes to JCQ guidance on AI in assessment
- Ofcom: younger phone owners, the rise of AI and consumption over creation (21 May 2026)
- Engadget: how to set up ChatGPT's parental controls to protect your teen (6 September 2026)
- TechRadar: here's how ChatGPT parental controls will work (2 September 2025)
- Anthropic: consumer terms of service (effective 8 October 2025)
AI helps with research and drafting. Dan checks the sources and is responsible for what is published.
How we check