Parents ask it, teachers ask it, and students ask it quietly: is this cheating? The honest answer is that "AI" is not one act. Using it to understand a topic and using it to produce the thing you are marked on are different acts that happen to use the same app. This guide separates them.

Two different questions

There are two questions and they get mixed together.

  1. Does this build learning? That is about the learner. A use can be fully allowed and still teach nothing.
  2. Is this allowed here? That is about the rules. A use can build learning and still break a task's rules.

A use can pass one test and fail the other. A learner who has AI write an essay in a task that bans it has broken a rule. A learner who has AI write the essay in a task that allows it has kept the rule, and may still have learned very little. We keep the two questions apart below, and only the second one is for the school to answer.

Five uses, from most to least learning

Five ways to use AI on homework, from most to least learning

  • ExplainBuilds learningAsk for an idea to be explained another way, then close the chat and try the problem.
  • PractiseBuilds learningAsk for practice questions on a topic and answer them from memory before looking.
  • CheckDepends on howUsually fine after a real attempt, and only learning if the learner fixes the mistake. For assessed work, ask first and acknowledge any use.
  • DraftDepends on howAllowed in some tasks and banned in others. Read the task's rules before using it.
  • SubmitReplaces learningHanding in AI-written work as your own is treated as misconduct by the policies we read.
The label is about learning first. Whether a use is allowed is a separate question that only the teacher, school or exam board can answer.

Explain. Asking AI to explain an idea a second way is like asking a patient friend. It builds learning as long as the learner then tries the problem themselves. It rarely raises integrity concerns.

Practise. Asking for practice questions and answering them from memory is one of the strongest ways to learn, which we cover in Best AI for studying. It is also hard to call cheating, since no assessed work is produced.

Check. Having AI look at finished work and point at errors can help, provided the learner made a real attempt first and fixes the mistakes themselves. If the tool rewrites the work and the learner pastes it back, it has moved from checking into drafting.

Draft. Here the answer is "it depends on the task". A teacher may allow AI for brainstorming or an outline and ban it for the final text. The risk is that a draft quietly becomes the work. If the learner could not explain the draft aloud to the teacher, it is not yet theirs.

Submit. Handing in AI-written work as your own is what the policies we read treat as misconduct. It is also the use where the learner gains least, because the learner's own thinking is missing. A related warning comes from a high-school maths trial, where students who practised with an unrestricted chatbot scored 17% below peers who never had AI once it was taken away (that trial studied practice, not submitting work) (Bastani et al., PNAS, 2025).

What schools and exam boards say

Policies differ, so read your own. A few public examples show the shape.

Scotland's exam body, SQA, says that where the use of generative AI is permitted, its use must always be acknowledged. It states that submitting generative AI outputs as a learner's own work is considered malpractice, and that assessed work must reflect the learner's own knowledge, skills and understanding (SQA, Generative AI in assessments).

England's Department for Education tells schools that teachers and leaders must use professional judgement, that settings should comply with the age restrictions set by AI tools, and that pupils "should only be using generative AI in education settings with appropriate safeguards in place, such as close supervision and the use of tools with safety and filtering and monitoring features". It points readers to the Joint Council for Qualifications' guidance on AI misuse and detection (DfE, Generative AI in education).

Rules in other systems, such as Malaysia's SPM, India's CBSE, Singapore's MOE or the IB, are set by those bodies and their schools. We have not summarised them here, so ask the school which rules apply to the task in front of the learner.

The pattern across the sources we could read is consistent: acknowledge any permitted use, never submit AI output as your own, and let the task's rules decide.

Three questions before you submit

Three questions before you submit

  1. 1

    Does the task say what is allowed?

    A task, school policy or exam-board rule beats anything a tool says.

  2. 2

    Is the work mine, in my words, from my understanding?

    If you could not explain it aloud to the teacher, it is not yet yours.

  3. 3

    Did I say what help I used?

    Where a use is allowed, saying so keeps it honest. Saying so does not make a banned use allowed.

If any answer is no, stop and ask the teacher before handing the work in.

The second question is the one that protects the learner. If they can explain the work aloud, answer follow-up questions on it and rewrite a paragraph of it on the spot, the work is theirs in the way that counts. If they cannot, the problem is not only a rule. It is that the learning did not happen.

For parents: what to say and what to watch

Start from curiosity: ask to be shown how it is used.

Start from curiosity, not accusation. Ask your child what they use AI for and have them show you. Then use the five uses above as a shared language. "Is that explaining or submitting?" is easier to say than "are you cheating?".

Agree one rule at home that matches the school's: the learner tries first, uses AI for explaining, practising and checking, and does not paste in anything that will be marked unless the teacher has said it is allowed. If the school has no clear policy, ask for one. Practical house rules by age are in AI homework helper: which kind helps.

For teachers: say it in the task

One line in the task saves a dozen arguments later.

In our view, most accidental misuse comes from silence. A learner who is told nothing will guess. Add one line to each task: what is allowed (for example explaining and practice), what must be acknowledged (for example a brainstorm), and what is not allowed (for example generated text in the submission). A task that asks pupils to show their working, or to talk through it, makes the question easier to answer without any software. If a piece of work worries you, do not rely on a detector alone. We explain why in AI detectors and students.

For older students: protect yourself

Keep your own notes and drafts. If you used AI where it was allowed, write down what you asked and what you changed. If you are unsure whether a use is allowed, ask before you submit. A teacher is far more likely to be helpful to the student who asked than to one who has to explain after.

Frequently asked questions

Is using ChatGPT for homework cheating? Not automatically. It depends on the task's rules and on what the learner does with it. Explaining and practising are usually fine, and so is checking after a real attempt, unless the work counts towards a qualification, in which case ask first and acknowledge any use. Handing in its output as your own is not.

Can I use AI to check my homework? Usually, once you have made your own attempt and unless the task says otherwise. For work that counts towards a qualification, ask first and acknowledge any use. Fix the errors yourself. If the tool rewrites the work and you paste it in, that is drafting, and the task's rules for drafting apply.

Will my teacher know? Do not rely on being undetected, and do not rely on detectors being reliable either. Teachers also notice a sudden change in style or knowledge, and may ask you to explain your work. The safer plan is not to submit what you cannot explain.

What counts as acknowledging AI use? Where AI is allowed, say what tool you used, what you asked it, and what you changed. Follow the format your teacher or exam board gives.

My child was accused of using AI. What now? Stay calm and ask what evidence exists beyond a detector score. Gather drafts and notes. We set out the steps in AI detectors and students.

Sources

  1. SQA, "Generative artificial intelligence (AI) in assessments": https://www.sqa.org.uk/sqa/107607.html
  2. UK Department for Education, "Generative artificial intelligence (AI) in education": https://www.gov.uk/government/publications/generative-artificial-intelligence-in-education/generative-artificial-intelligence-ai-in-education
  3. Bastani et al., PNAS 122(26), 2025 (abstract via RePEc): https://ideas.repec.org/a/nas/journl/v122y2025pe2422633122.html