What should I actually say to my child about AI?
Three questions before they share anything AI helped make: Is it fair? Is it safe? Is it honest? They are short enough to remember and they cover most of what goes wrong. Beyond that, the most useful thing a parent can do is ask to see the conversation, not just the output. The prompt tells you far more about what your child was thinking than the answer does.
Free to read. No account, nothing to sign up for.
What we teach: a 3-question ethics check, used before anything gets shared.
Three questions worth memorizing
Is it fair? Is it safe? Is it honest? Asked before anything AI helped make gets shared with anybody.
Fair covers whether it treats people properly and whether using it here gives an unearned advantage over someone else. Safe covers what went in as much as what came out. A child pasting personal details into a chat box is the common failure, not the dramatic one. Honest covers whether anyone reading this would be surprised to learn how it was made.
The reason it is three short questions and not a policy is that a rule only works if it can be recalled at the moment it is needed, which is usually late, alone, and under time pressure. Anything longer does not survive that moment.
Ask for the conversation, not the output
The single most useful habit available to a parent: ask to see the prompt, not the result. What your child asked for reveals what they were thinking, where they were stuck, and what they believed the task was. The output reveals what a machine produced.
It also changes the emotional register completely. "Let me see what you asked it" is curiosity. "Did you use AI for this?" is an accusation, and it reliably produces either a defensive no or a shutdown, neither of which tells you anything.
Say the parts you do not know
You do not need to understand this well to be useful here, and pretending to will cost you more than admitting you do not. "I am not sure whether that is true, how would we find out?" is a better contribution than a confident wrong explanation, and it models the exact behavior you want.
Children are quick to notice when an adult is bluffing about technology, and the conclusion they draw is that you are not a person to bring this to. That is a worse outcome than any individual misunderstanding.
Openers that work better than a lecture
"Show me something it got wrong." Nearly every regular user has an example, it is fun to share, and it establishes fallibility without you having to assert it.
"What would you never use it for?" This asks a child to state their own limits, which tends to produce more thoughtful answers than being handed limits, and tells you where they actually are.
"What did you change?" The most useful follow-up about any piece of work. A student who changed things was thinking; a student who changed nothing has something worth talking about.
"Would you tell your teacher you used it?" The honesty test, put in a form a child can answer immediately. Hesitation is the informative part, and it is usually visible before they say anything.
Where this is taught
IF YOU WANT TO GO FURTHER
This answer comes from the program we teach: sixteen video lessons that take a student from “what even is this” to directing AI on purpose. Parents get their own access alongside their child, because it is hard to guide something you have not seen.
The first lesson is free to watch, no account needed. Start there before you decide anything.