Please note: this guide is general information for parents, not medical, psychological or other professional advice, and it can’t take your child’s individual situation into account. If you’re worried about your child’s health, wellbeing or safety, speak to their school or a qualified professional such as your family doctor. In an emergency, contact your local emergency number.
Children follow rules better when they help write them and when the rules are short enough to remember. Below is a family AI agreement for homework you can copy, edit together and stick on the fridge — with versions by age, a list of what never to share with a chatbot, and a simple way to check the rules are working.
Why an agreement works better than a ban
AI tools are built into phones, search engines, school platforms and homework apps. A total ban is hard to enforce and tends to push use out of sight, where you can’t help your child use it well. An agreement does three things a ban can’t: it names the good uses, it marks clear no-go zones, and it gives you a routine for checking in.
The rules below follow one principle: AI can be a tutor, never a ghostwriter. It may explain, question, hint and quiz. It may not produce work your child hands in as their own.
The eight core rules
- Try first. Have a real go before asking AI. Struggle a little — that’s where learning happens.
- Tutor, not ghostwriter. AI can explain, give hints and quiz you. It doesn’t write, solve or answer for you.
- Set it up as a tutor. Use the family tutor prompt at the start of every homework chat.
- School rules win. If a teacher says no AI for a task, it’s no AI.
- Explain it back. If you used AI, you can explain your work without the screen. If you can’t, redo it.
- Check what it says. AI can be wrong while sounding certain. Check important facts in your textbook or with your teacher.
- Keep private things private. No full names, addresses, school names, passwords or photos of people.
- Tell us if something feels wrong. If an AI says something upsetting, strange or inappropriate, close it and tell a parent. You won’t be in trouble.
The template: copy, edit, sign
Print it, or copy it into a document and edit it together. The blanks are there on purpose — filling them in as a family is what makes it yours.
OUR FAMILY AI AGREEMENT — HOMEWORK What AI is for in our house: - Explaining things in a different way when we're stuck - Giving hints after we've tried - Quizzing us before tests - Feedback on work we've already written (without rewriting it) What AI is NOT for: - Writing essays, paragraphs, stories or answers we hand in - Solving maths or science problems for us to copy - Tests, quizzes or anything the teacher says is AI-free - [ADD YOUR OWN] Our rules: 1. Try first. Ask AI only after a real attempt. 2. Start every homework chat with our tutor prompt. 3. School rules always win. 4. If we used AI, we can explain our work without the screen. 5. We check important facts — AI can be wrong. 6. We never share full names, addresses, school names, passwords or photos of people. 7. If an AI says something upsetting or strange, we close it and tell a parent. No trouble for telling. Where and when: [e.g. kitchen table only / not after 8 PM / parent in the room for under-13s] Which tools: [e.g. ChatGPT on Mum's account, Gemini on the family laptop] How we'll check in: [e.g. Sunday chat, 5 minutes, look at one AI conversation together] If a rule is broken: [e.g. redo the work without AI; AI paused for a week; talk about what went wrong] We'll review this agreement on: [DATE] Signed: ______________ ______________ ______________
Rule 2 refers to a tutor prompt. You’ll find a ready-made one in our tutor prompt guide.
Rules by age
Under 9
- No chatbot accounts of their own. The major tools set minimum ages of 13 or above for personal accounts in most countries (details in our age rules guide).
- AI happens on a parent’s device, with a parent typing and reading aloud.
- Most of the value at this age is for the parent: understanding the topic before helping.
Ages 9–12
- Still on a parent’s account if under the tool’s minimum age. A parent sets the tutor prompt and stays in the room.
- Short sessions: one problem or one concept at a time.
- Explain-back after every session.
Teenagers
- Their own account if they meet the tool’s age rules, with your permission as the terms require.
- More independence, clearer no-go zones: no AI-written sentences in submitted work, ever.
- Regular check-ins rather than constant supervision. Ask to see one conversation a week — not to catch them out, but to talk about how they’re using it.
- ChatGPT offers parental controls for linked teen accounts, including quiet hours. They don’t let you read conversations, so they complement an agreement rather than replace it.
What never to share with an AI chatbot
Conversations with AI tools may be stored and, depending on the service and settings, used to improve the models. Treat a chatbot like a public space.
- Full name, home address, phone number or school name
- Passwords, account details or anything financial
- Photos of themselves, friends or family
- Private details about other people
- Health information or personal problems — those belong with a trusted adult
When you photograph a worksheet, crop out names and school logos first.
AI-free zones
Some work should stay entirely human. Common choices:
- Tests and quizzes, including take-home ones, unless the teacher says otherwise.
- First drafts. Ideas and sentences come from your child; AI feedback, if allowed, comes after.
- Reading. AI summaries don’t replace reading the book.
- Practice that builds speed, such as times tables and spelling — although AI can usefully quiz on them.
Making it stick: the weekly check-in
Rules that nobody checks fade within a fortnight. A five-minute weekly check-in keeps them alive:
- Ask: “What did you use AI for this week?”
- Look at one conversation together. What worked? Did it ever give an answer when it shouldn’t have?
- Pick one thing they learned and ask them to explain it to you.
- Ask if any rule feels unfair or unclear, and adjust together.
Rule 3 is easier when every situation already has a ready-made prompt. The AI Tutor Vault has 40 of them, all built to explain and hint rather than answer. See the AI Tutor Vault.
When a rule is broken
It will happen. Respond in proportion and keep the focus on learning:
- First time, minor: redo the task without AI, and talk about why the rule exists.
- Repeated: AI use only with a parent present for a set period.
- Submitted AI work at school: support your child to put it right with the teacher, following the school’s process. Our guide to AI and cheating covers this conversation.
Also ask why. A child who used AI to write an essay may be overwhelmed, behind, or anxious — and that needs solving too.
Frequently asked questions
How strict should our AI rules be?
Strict about the no-go zones (submitted work, tests, privacy), flexible about the rest. Overly detailed rules are hard to follow. Eight short rules your child helped write will beat thirty they didn’t.
Should I read my teen’s AI conversations?
Agree on it in advance rather than checking secretly. A weekly look at one conversation, done together, builds trust and keeps the rules visible. ChatGPT’s parental controls do not show you conversation content.
What if school encourages AI use?
Then follow the school’s guidance for schoolwork and keep your agreement for everything else. Many schools teach AI for brainstorming or feedback; your family rules can match that.
When should we review the agreement?
Every school term, or whenever your child starts using a new tool or moves to a new school stage. Children’s needs and the tools both change quickly.
Next steps
Set aside fifteen minutes this weekend. Read the eight rules together, fill in the blanks in the template, and sign it. Then set up the tutor prompt your rule 2 refers to, using our step-by-step guide to using ChatGPT for homework without cheating.