Is ChatGPT safe for kids? Age rules and the settings to change

ChatGPT and other chatbots are made for adults and teens. That does not mean children never come across them, so it is worth knowing the rules and the risks.

Age rules

OpenAI's terms say users must be at least 13, and those under 18 need a parent's permission. Other services have their own limits, and some offer special accounts for children managed through a parent account. Rules change, so check the current terms for any tool your child uses.

What can go wrong

• Wrong answers. Chatbots can sound sure and be wrong, and they sometimes make up facts and sources.
• Unsuitable content. Filters are not perfect.
• Privacy. Children may share names, schools, photos or feelings without realising where that information goes.
• Doing the work. A chatbot can write a child's homework, which means they do not learn.
• Over-trust. Because chatbots answer kindly and instantly, young children can treat them as friends.

Safer ways to use a chatbot with a child

1. Sit with them while they use it, especially for the first weeks.
2. Use your account and look at the history together.
3. Check the privacy and data settings in the account, and switch off options to use chats for training where you can.
4. Set the purpose. Ask for ideas or explanations, not finished homework.
5. Check anything important against a second source.

Family rules to agree on

• No real names, addresses, schools or photos.
• A chatbot is a tool, not a friend or a doctor.
• If it says something strange or upsetting, close it and tell a grown-up.
• We check before we believe.

Our AI safety rules every kid should know turns these into a list for the fridge, and the free mini course includes a chatbot agreement to sign.

For younger children

Under 13, it is usually better to learn about AI without chatting with it. Games, cards and lessons teach the ideas without the risks. See 12 AI activities for kids.