AI homework rules for families: a house agreement you can print
Write the rules down before there's a conflict. Use the four permission levels per assignment type, require your child to disclose AI use rather than hide it, and agree in advance on what happens when a rule is broken. Sit down together and adjust the wording — a rule they helped write is one they'll follow.
Why this needs to be written down
"Don't cheat with AI" is not a rule. It is a sentiment. Your child will interpret it in whatever way makes the next two hours easier, and you will find out about the interpretation gap during an argument.
A written agreement removes the ambiguity and — more usefully — removes you from the position of arbitrary authority. The rule is the rule. You are both looking at the same piece of paper.
Read it together. Let them push back on the wording. A rule your child argued about and then agreed to is enforced from the inside.
The agreement
1. AI is allowed in this house
We are not pretending it does not exist. Learning to use it well is part of your education. These rules exist so it helps you instead of replacing you.
2. Every assignment has a level
Before you start, decide which level applies. If you are not sure, ask.
- Level 0 — AI off. Spelling, times tables, reading logs, memorization, anything where the repetition is the point.
- Level 1 — AI explains, never answers. You may ask it to teach you the concept. You may not show it the actual question.
- Level 2 — AI on the scaffolding. Brainstorming, self-quizzing, checking your own writing for grammar. The ideas and sentences stay yours.
- Level 3 — AI as a partner, disclosed. Research, coding, presentations. You must be able to say exactly what it did and what you did.
3. Nothing AI wrote gets submitted as yours
Not a sentence, not a paragraph, not a "cleaned up" version of your paragraph that came back as a different paragraph. If the words are the model's, they do not go in the box with your name on it.
4. You have to be able to explain it
Any completed homework can be explained out loud, without the screen, in your own words. This is the real test and it applies to everything. If you cannot explain it, it is not finished.
5. Always check the answer
AI is confidently wrong on a regular basis. Treat every answer as a claim from a clever stranger who has not read the textbook. Verify anything that matters.
6. Telling us is never the thing you get in trouble for
If you used AI more than you should have, say so. The consequence for telling us is that we redo the work together. The consequence for hiding it is bigger, because hiding it is the actual problem.
7. School rules override house rules
If a teacher says no AI, the answer is no AI, even if this agreement would allow it. When a teacher's rule is unclear, we ask them — we do not guess in our favor.
8. Keep your work
Draft in a document with version history. Keep your outlines and notes. AI detectors flag honest work sometimes, and your process is the evidence that protects you.
When a rule gets broken
Agree on this part in advance, while nobody is upset. A workable default: the assignment gets redone at level 1, together, and the level for that subject drops one step for two weeks. No lecture, no device confiscation — the consequence is doing the thinking that got skipped.
Reviewing it
Try it on this one
Check the homework in front of you.
Paste the assignment — or a screenshot of the AI chat — and get a plain-language read on whose thinking it is.
Revisit the agreement every school year, and any time your child moves up a level. A rule set written for a 10-year-old is insulting to a 14-year-old, and one written for a 14-year-old is useless for a 10-year-old.
For the reasoning behind each level and how to run the explain-it-back test, read our parent's guide to AI homework help. For the school-policy side — what counts as cheating and what to do about a false accusation — read is using AI for homework cheating?.
Frequently asked questions
What are good AI rules for kids doing homework?
Set a permission level per assignment rather than one blanket rule: AI off for memorization and recall practice, AI as an explainer for concepts, AI for scaffolding like brainstorming and self-quizzing, and AI as a disclosed collaborator for projects. Add a non-negotiable that nothing AI wrote is submitted as the child's own work.
Should AI rules be different for different ages?
Yes. Younger children should mostly stay at supervised, explanation-only use, while teenagers can handle disclosed collaboration on projects. Review the agreement each school year — rules written for a 10-year-old do not work for a 14-year-old.
What should happen if my child breaks the AI rules?
Agree on the consequence in advance, while nobody is upset. A workable default is redoing the assignment together with AI limited to explanations, and dropping that subject's permission level for two weeks. Make disclosure safe so hiding is never the better option for them.
Do house AI rules override the school's rules?
No — school rules always win. Your household agreement should state this explicitly, and should tell your child to ask the teacher when a classroom rule is unclear rather than assuming the more permissive reading.
Have a specific piece of homework in mind?
Paste it — or a screenshot of it — into the Homework Check on our homepage and ask your question. You'll get a plain-language read on whether the thinking looks like your kid's, what stood out, and the questions worth asking them. First check is free.