AI and Exam Rules: What's Actually Allowed (UK Guide)
Could using AI get a grade taken away? For exams the answer is simple; for coursework it's nuanced. Here's exactly where the exam boards draw the line.
Here's the anxiety in one sentence: could using AI to help with schoolwork get your child's grade taken away? For written exams the answer is simple. For coursework it's more nuanced — but the exam boards have set out clear rules, and once you understand the one core distinction, the rest follows.
This is a plain-English guide to where the line sits. It's based on the JCQ guidance “AI Use in Assessments” (the Joint Council for Qualifications speaks for the main UK exam boards including AQA, Edexcel, OCR, WJEC and CCEA). Always check your child's school policy and the current JCQ document too, as details are updated — links are at the end.
In an exam: no AI — you're in closed conditions with no devices. In coursework / non-exam assessment (NEA): AI may be used to help you learn and research, but any AI content you use must be acknowledged, and work you submit must genuinely be your own.
Written exams: no AI, full stop
This one is straightforward. Exams are sat under supervised, closed-book conditions with no internet and no devices, so AI simply isn't available — and attempting to smuggle it in is serious malpractice, in the same category as any other prohibited material. There's no grey area here.
Coursework and NEA: allowed, but with rules
This is where most of the worry — and most of the nuance — lives. Under JCQ's guidance, students may use AI tools during the research and preparation stages of many coursework and non-exam assessments. What they must not do is pass off AI-generated material as their own work. Specifically, JCQ identifies AI misuse as things like:
- copying or paraphrasing AI-generated content so that what's submitted is no longer the student's own work;
- using AI to write part or all of an assessment and not disclosing it;
- failing to acknowledge AI tools where they've been used as a source;
- submitting work with a false declaration of authenticity.
Crucially, students can't be awarded marks for any part of the work that relies on unacknowledged or improperly acknowledged AI — because they haven't independently met the marking criteria. Note too that some subjects and boards restrict AI use more tightly than others, so the safest first step is always to check the specific rules for that qualification.
How to acknowledge AI properly
If a student does use AI as a source in coursework, JCQ expects them to be transparent about it. In practice that means:
- Name the tool and the date — for example, “ChatGPT, 20 January 2026”.
- Keep non-editable copies (such as screenshots) of the prompts used and the AI's responses.
- Briefly explain how it was used, and reference it like any other source.
- Independently verify anything the AI produced, and cite the real, checked sources.
- Submit these records alongside the work so the teacher can authenticate it.
The teacher must be able to sign that the work is the student's own — so anything that makes that impossible to verify is a problem.
What happens if the line is crossed
Malpractice is taken seriously. Depending on severity, sanctions can range from losing the marks for the affected part of the work, up to disqualification from the whole qualification and, in the most serious cases, being barred from exams for a period. A single misjudged shortcut is genuinely not worth that risk — which is the strongest argument for using AI the right way rather than avoiding it out of fear.
A sensible rule of thumb
Use AI to help you understand and improve your own work — never to produce the work you hand in.
If AI helped you think — explaining a concept, quizzing you, giving feedback on your own draft — that's revision, and it's fine (acknowledge it where required). If AI wrote what you submitted, that's a problem, however much you reword it. Our companion guide shows exactly how to stay on the right side of this: using AI for coursework without cheating.
Questions students and parents ask
Can AI detectors prove my child cheated?
No — AI detectors are unreliable and produce false positives, and JCQ's process doesn't rest on them. Authenticity is judged on evidence the teacher can verify (drafts, process, the ability to discuss the work), which is another reason to keep your working. We cover wrongful accusations in our guide “Is my child cheating with AI?”
Is using AI to check spelling or grammar allowed?
Basic spell-check and grammar tools are generally fine and long-established, but heavy AI rewriting of your text is a different matter and can stray into misconduct. When in doubt, keep the writing yours and check the rules for that specific assessment.
Do these rules apply to A-Levels and BTECs too?
JCQ's guidance covers its member awarding bodies across GCSE, A-Level and many vocational qualifications. The core principle — acknowledge AI, submit only your own work — applies broadly, but always confirm the exact rules for the specific qualification and board.
References & further reading
This guide is general information, not a substitute for your school’s policy or the current official JCQ guidance, which is periodically updated. Always check both before you act.
- JCQ (2025). AI Use in Assessments: Protecting the Integrity of Qualifications. Joint Council for Qualifications. jcq.org.uk
- JCQ (2025). Instructions for Conducting Coursework 2025–26. Joint Council for Qualifications.
- Your own school’s AI and assessment policy, and current guidance from your exam board (AQA, Pearson Edexcel, OCR, WJEC/Eduqas, CCEA).
The method behind the games and guides
The Velvet Method is the six-step system these guides all point back to — taught step by step, with a full prompt library, for GCSE, A-Level and IBDP.
Explore the Velvet Method course →