A family AI agreement template you can print and sign tonight

Why a written agreement beats a lecture

Most households handle AI the same way: an anxious conversation after something goes wrong, then nothing until the next incident. The rules exist only in the parent's head, which means every enforcement feels arbitrary and every negotiation starts from zero.

A written agreement fixes three things at once. It makes the rule visible before the moment of temptation. It makes it mutual — you sign too, which changes the emotional register completely. And it makes disputes boring: you point at line four instead of arguing about intent.

What goes in it — the child's five promises

Keep the promises behavioral. "Use AI responsibly" is unenforceable. These are the five that hold up:

• **I'll tell you when I use it.** Not permission — disclosure. A one-line note of what the AI did is enough.
• **I won't hand in anything I can't explain.** The test isn't who typed it, it's whether the child can defend paragraph three out loud.
• **I'll check the facts it gives me.** Every number, name and date gets a second source.
• **I won't put personal details into it.** No full name, school, address, photos, or anyone else's information.
• **I'll tell you if something feels wrong.** Anything upsetting, adult, or manipulative — no punishment for reporting it.

Imagine you are filling out the agreement beside a real assignment. It says to explain a character's decision using evidence from the book. Ask which help would be allowed: explaining an unfamiliar instruction, suggesting a complete answer, or correcting a sentence the child wrote. Check the teacher's rule before settling the example. Writing down one allowed request and one prohibited request makes the promise easier to use on a rushed evening.

Suppose the privacy promise feels vague. Use an invented worksheet with a name and address at the top, and ask what information an explanation of fractions actually needs. Cross out details that identify a person. You can practice the decision without exposing a real child's information. The aim is to recognize unnecessary details before submitting anything, not to memorize a list while already rushing through a task.

For the promise to check answers, imagine a suggested fact appears in a history note. Ask where your child would verify it and how they would record the source. Then make that check together. 'I'll be careful' becomes a visible action: find the claim, compare it with a reliable reference, and correct or leave out what cannot be supported. A tool's certainty doesn't replace the source.

The three promises you make back

This is the half most templates skip, and it's the half that makes kids sign willingly.

• **I won't punish you for telling me the truth about what you used.** Consequences apply to hiding, not to disclosing.
• **I'll learn the tools with you instead of banning what I don't understand.**
• **I'll ask what you think before I change a rule.** Rules get revisited as your child gets older, not imposed permanently at age nine.

Then leave two blank lines for house-specific rules — screen-free hours, which subjects are off limits, whether the laptop lives in the kitchen. Families follow rules they wrote themselves. For what's age-appropriate to allow in the first place, see our AI homework rules for families and our guide to how old kids should be to use ChatGPT.

Suppose your child tells you they used a generated opening sentence when the class didn't permit it. Follow the agreement about honest disclosure while addressing the assignment itself. Ask what made starting difficult, then help them plan an independent replacement under the teacher's rules. Being truthful doesn't make the original use acceptable, but it lets the family work with an accurate account instead of guessing from a finished page.

Imagine you realize your own first response was sharper than intended. You can say that, return to the written agreement, and separate the rule from your frustration. The parent promises belong on the page too. If the family needs to revise a consequence, discuss it at the next calm review rather than inventing a new one halfway through an argument.

How to run the conversation

Do it at the table with the page printed, not on a screen and not in the middle of a homework argument. Read each line out loud and ask your child whether it's fair. Expect pushback on the disclosure line — that's the one that matters most, so trade for it: they disclose, you don't punish disclosure.

Sign it, date it, and stick it where the homework happens. Re-sign it at the start of each school year; the rules that fit a fourth-grader don't fit a ninth-grader.

Suppose the first review happens after two ordinary homework weeks. Bring one situation where the agreement was clear and one where it wasn't. Maybe a request for spelling help turned into a proposed rewrite. Ask where the boundary should have been noticed and which wording would make it clearer next time. Keep teacher rules above the household permission. The review can improve the agreement without turning an exception into automatic permission for future assignments.

What to do when it gets broken

It will. The point of the agreement isn't perfect compliance — it's giving you a calm script for the moment after. Go back to the page, find the line, ask what happened, and rewrite the line together if it turned out to be unrealistic. A rule that's been renegotiated once is more durable than one that's been broken silently five times.

And if the reason you're here is a specific piece of homework that doesn't look like your kid's writing, start with is using AI for homework cheating? before you start with the accusation.