Whether AI homework help counts as cheating depends on what the tool does and what your child's school allows: handing in AI-made work as their own is a problem, while being questioned or quizzed by an AI is not. The most useful test for any tool is whether your child could do a similar problem alone afterwards.
Most families now have an AI tool somewhere in the house. This article explains the two very different kinds, what schools worry about, and gives you a checklist you can apply to any app.
Two kinds of AI homework helper
The answer-giver. Photograph a question, or paste it in, and get a finished solution or a finished paragraph. It is fast and often right. It is also the thing most likely to produce work that looks like your child's but is not.
The tutor. It asks what they have tried, narrows the problem, names the method, and makes your child do the last step. It feels slower, sometimes irritatingly so, because the thinking has moved back to your child.
The same underlying technology can power either. What differs is how the product is designed and what it refuses to do. That is why "is AI good or bad for homework?" is the wrong question. The right question is what this particular tool does with the child's effort.
What schools worry about
Schools generally worry about two things.
Whose work is it? The Department for Education's policy paper on generative AI in education says schools and colleges may wish to review their homework policies, and other unsupervised study, to account for generative AI, including guidance on when using AI tools is acceptable. For qualifications, the exam boards' joint body JCQ says work must be the student's own, and misuse of AI to the point where work is not theirs can count as malpractice. KS3 is not yet that stage, but habits form early.
Over-reliance. The DfE's product safety standards for AI in education name "cognitive deskilling" as a harm products should try to prevent, and expect learner-facing tools not to give final answers or full solutions by default but to start with hints. A child who always gets the answer does not build the muscle for exams, where there is no AI in the room.
Each school sets its own rules on AI for homework, and they vary by school. Look for the school's policy, or email the form tutor.
Accuracy: AI can be confidently wrong
AI chat tools can produce fluent answers that are wrong, particularly on multi-step maths, dates, quotations and sources. A wrong answer delivered in a confident tone is harder for a 12-year-old to question than a wrong answer from a classmate.
Teach one habit: for anything that will be handed in, check it against the textbook, class notes or the teacher. Better still, ask the tool to explain its reasoning step by step and see whether your child can follow and challenge it.
Privacy and safety: questions to ask
The DfE has published product safety standards for generative AI in education, covering areas such as content filtering, data protection and learners' wellbeing, and says schools and colleges may find them helpful in judging which products are safe. You do not need to read it all, but its themes are good questions for any app:
- What is the minimum age, and is it enforced or just stated?
- What data is stored, for how long, and can it be deleted?
- Are children's conversations used to train AI models?
- What happens if a child says something worrying, such as being upset or unsafe? Does the product point to real help?
- Does it use nudges to keep children on the app, such as streak-loss warnings?
- Can a child reach purchases without a parent?
The Information Commissioner's Office Children's Code (its age appropriate design code) sets standards for online services likely to be used by children; a good product will be able to say how it treats it.
A parent checklist for any AI study tool
- Does it make my child think? Watch one session. Do they do the last step?
- Can my child explain the answer afterwards? Ask them to teach you.
- Is it clear about its limits? Good tools say what they will not do.
- Do I know what data it holds? Read the privacy page for ten minutes.
- Does it match the school's rules? Check the AI policy.
- Is there a human route for worries? A fixed signpost to a parent or Childline is a good sign.
- Does it fit a routine? Short, specific sessions beat open-ended scrolling.
A script to get started: "Show me how you used it tonight, and explain one answer to me as if I were Year 6."
- Watch one sessionDoes my child think?
- Ask them to teach youCan they explain it?
- Check the dataWhat is stored, for how long
- Check school rulesVaries by school
- Human route for worriesA signpost to a parent or Childline
- Short, specific sessionsNot open-ended scrolling
Where Heddy fits
Heddy is one example of the tutor design. It is an app for students aged 13 and over in UK schools, and its tutors climb a four-rung hint ladder whose top rung is a worked similar problem, never the answer to your child's own question. It also has a fixed safeguarding line pointing to a parent or Childline 0800 1111; the safety guide covers what it stores. Whatever tool you choose, the checklist above is the judge, not the marketing. For the wider habit, see how to help with homework without giving the answer.
Updated 30 Sep 2026. Written by Duke Harewood, who builds the tutors at aitutors.me.
Questions people ask
Is using AI for homework cheating?
It depends on what the AI does and what the school allows. Having AI produce work a child hands in as their own is a problem; using it to explain, quiz or question them is different. Check your school's AI policy, because rules vary by school.
What is the difference between an AI answer-giver and an AI tutor?
An answer-giver returns the solution. A tutor asks questions, gives graded hints and makes the child do the thinking. The test is simple: after the session, could your child do a similar question alone?
Is an AI tutor safe for children in the UK?
It depends on the product. Ask about age limits, what data is stored and for how long, whether conversations are used to train models, and what happens if a child says something worrying. The Department for Education has published product safety standards for generative AI in education.
Can AI make mistakes with homework?
Yes. AI tools can state wrong answers confidently, so children should check important facts and methods against their textbook or teacher.