Schools protect academic integrity by designing assessment so that what counts is done where AI isn’t available or its use is visible: supervised assessments, coursework drafted under supervision, and conversations about work. Cambridge puts “the primary responsibility for identifying any inappropriate use of generative AI” on teachers who know the students best. Detection software alone is not the answer.
What do the exam boards require?
Clear rules, with teachers responsible for applying them:
| Setting | Rule | Source |
|---|---|---|
| Supervised exams | Students “must not be able to use” AI tools | JCQ, April 2025 |
| Cambridge coursework | AI only for initial research or brief, critically discussed quotation of about 50 words; “candidates must not use online writing assistant tools” | Cambridge policy; Handbook 2026 |
| Pearson coursework | AI must be “clearly referenced”; teacher and learner sign a declaration | Pearson AI guidance |
| Authentication | Teachers must not authenticate work they suspect used AI inappropriately | Cambridge policy |
Sources: Cambridge generative AI policy, Cambridge Handbook 2026, Pearson, JCQ. Full detail is in AI in IGCSE coursework.
Why isn’t AI detection enough?
Because the boards themselves rely on teacher judgement, not a detector score. Cambridge lists warning signs rather than a tool. They include unusually accurate spelling and grammar, Americanised spelling, heavily Latinate vocabulary and “a lack of any references after 2021”. Cambridge advises schools to “keep drafts and plans of coursework under supervision”. Pearson suggests examining work “at intermediate stages”, and says a verbal presentation “may be helpful” where AI use is suspected. None of these rests on software. A school that leans on detection risks two problems: missing AI use it cannot see, and wrongly accusing students on a score it cannot explain.
Does AI use actually harm learning?
It can make work look better without making students better. The OECD’s Digital Education Outlook 2026 warns that generative AI “may enhance the apparent quality of student work … without improving their actual learning”. It cites a randomised trial in Türkiye in which students’ gains from using AI vanished “when their knowledge was assessed in a closed-book environment” (OECD). For IGCSE and A Level schools, that is the real integrity risk. Homework done with AI can hide a gap that only shows in the exam hall.
How should schools design assessment that still works?
Match each assessment to what it is for, and make the evidence that counts visible.
- Supervised assessment for anything that counts. Mocks, end-of-unit tests and evidence for predicted grades should be sat in class, on paper or locked-down devices.
- Drafts under supervision for coursework. Plan, draft and revise in lessons, with dated drafts kept, as Cambridge advises.
- Short conversations. Ask students to explain how they reached an answer or argument. Pearson notes a verbal presentation “may be helpful”.
- Process over product in homework. Set tasks that show working, planning or reflection, or treat homework as practice that doesn’t count.
- Frequent low-stakes checks. Short in-class quizzes show what students can do without help, and they are good retrieval practice in their own right.
- Teach acceptable use. Students should know when AI is allowed, how to acknowledge it, and why unacknowledged use is malpractice.
What should a school’s AI integrity policy say?
Which uses are allowed for each type of task, how use is acknowledged, how work is authenticated, and what happens when misuse is suspected. Our AI policy template covers all four, aligned to Cambridge and Pearson. The Cambridge Handbook is clear on timing. If inappropriate AI use is found during the course, “you should resolve this internally”. If it is found at submission, “you must not submit the work”.
How schools do this with AI Buddy
AI Buddy supports the supervised, low-stakes side of this. Teachers can run in-class practice and mini-tests on past-paper-style questions, marked instantly, so they see what each student can do in the room. Because AI Buddy’s AI feedback sits inside practice, not coursework, it stays within the boards’ rules. It also gives teachers a record of each student’s independent performance to compare against homework or coursework that looks out of character.
Frequently asked questions
How can schools protect academic integrity when students use AI?
Make assessment that counts supervised or visible: in-class tests, supervised coursework drafts, and conversations about work. Teach students the acceptable uses of AI and how to acknowledge it.
How do schools deal with AI cheating?
By designing assessment where AI use is either impossible or visible, and following the boards’ authentication rules. Cambridge says not to submit work with inappropriate AI use found at submission.
Should schools use AI detection tools?
Not on their own. The exam boards rely on teacher judgement, supervised drafting and warning signs, not detector scores.
What is AI-proof assessment design?
Assessment where the evidence that counts is produced under supervision or explained by the student, with homework treated as practice rather than proof.
Discover how AI Buddy helps schools strengthen teaching, learning and evidence-informed school improvement. Or start a short consultation with our schools team using the form below — we will get back to you directly.
Sources
- Cambridge International, The use of generative AI in coursework and Cambridge Handbook 2026
- Pearson, Artificial Intelligence (AI) Guidance
- JCQ, AI Use in Assessments (April 2025)
- OECD, Digital Education Outlook 2026