Home / Parent Guide / Is ChatGPT safe for kids?
Is ChatGPT safe for kids?
A public chatbot can be useful, and it can also be the wrong tool for a child. The useful question is not “is AI good or bad?”. It is who is using it, how old they are, what they type into it, and whether an adult can see what comes back.
Start with the age rule, then check it again
OpenAI’s published terms have required users to be at least 13, with extra conditions for teenagers. Other chat products set their own minimums, and those terms change. Before a child uses any public chatbot, read the current age and parental-consent rule on that company’s site. A family subscription does not cancel a provider’s age limit.
In the GIofAI Kids program, children under 13 do not independently use open public chatbots, including ChatGPT. They work with supervised, approved tools such as a browser activity that trains a model from the laptop camera, or a teacher-managed Scratch project. An adult is in every breakout room. The current list is the Approved AI & Learning Tools register, and a tool can be withdrawn if its age or privacy terms change.
What actually goes wrong
The common problems are ordinary, which is why they are easy to miss.
- Confident mistakes. A chatbot can invent a source, a date or a school rule and sound sure about it. Children need a habit of checking, which is covered in how to tell when AI is making things up.
- Private information. A homework prompt can include a full name, a school, a photo, a medical detail or a friend’s message. Once it is typed into a consumer chat, you should assume it may be stored under that service’s terms.
- Schoolwork. Using AI to understand a question is different from pasting an answer into an assignment. See should kids use AI for homework?
- Open conversation. A general chatbot is not a closed classroom activity. It can wander into topics you did not choose, including advice a child should hear from you or from a professional.
A family rule that is specific enough to use
Three sentences are enough to start. The child does not use a public chatbot alone before 13. From 13, they use it where you can see the screen, and they do not type names, schools, photos or passwords. If the answer will be handed to a teacher, the child has to be able to explain it without looking at the screen.
That is also the shape of the Family AI Agreement learners write during Future AI Innovators and bring home. The point of the agreement is a rule the child helped write, not a poster on the fridge that nobody reads.
Safe does not mean unused
Avoiding every AI tool until 18 leaves a child to meet it later, usually through a classmate’s phone, with no practice at spotting a bad answer. Supervised practice is the middle path: a named activity, an approved tool, a time limit, and a conversation afterwards. Ten questions to ask tonight is a simple way to start that conversation.
Future AI Innovators is a live online AI course for kids and teenagers aged 8–17 across Australia, running on school-holiday mornings from 4 to 15 January 2027. No coding experience is required. The safety approach, the tool list and the fees are in the Parent Guide.