Schools are under pressure to respond to AI quickly. Staff see homework that feels oddly polished, hear rumours about chatbots writing essays, and worry that classroom rules are already out of date. In that atmosphere, it is easy to jump straight to bans, sanctions or broad warnings. Yet policy written from anxiety often misses the real picture. A short audit gives leaders something better: evidence from students themselves.
That matters because student AI use is rarely just one thing. In one form group, a pupil may use AI to plan revision flashcards. Another may ask a chatbot to explain a maths method after school. A third may use it to draft homework they do not fully understand. A fourth may use it for emotional reassurance late at night. These are not identical behaviours, and they should not trigger identical responses. If your school is reviewing expectations, it helps to pair this audit with a broader policy process such as January INSET AI policy sprint pack.
A 30-minute tutor-time audit is designed to reveal patterns, not culprits. It is not an investigation. It is a quick, low-stakes way to find out what students say they are using AI for, where they feel unsure, and where school guidance may be too vague to be useful.
Done well, it can surface several important truths. You may discover that most use is mundane rather than dramatic: summarising notes, checking definitions, generating quiz questions or planning revision timetables. You may also find pressure points. Students might admit they are unclear about what counts as acceptable help. They may say different teachers give different messages. Some may reveal unsafe or emotionally dependent use that belongs in a safeguarding conversation rather than an academic misconduct one. That is why this work sits alongside wider thinking about school safeguarding pre-flight checks for AI chatbots.
In practical terms, this audit should help you answer five questions: How common is AI use in everyday school life? What tasks are students using it for? Where are the grey areas causing confusion? Which uses appear helpful, ineffective or risky? And what support do students actually want from school?
Those questions are useful because they move the conversation away from abstract debate. Rather than asking, “Is AI good or bad?”, you are asking, “What is happening here, in our setting, among these students?” That gives pastoral leaders, classroom teachers and senior teams a firmer basis for action.
The survey should be anonymous, short and plain-speaking. Avoid legalistic wording. Avoid asking for names, classes or identifying examples. If students think the form is really a trap, your data will be poor.
Here are ten questions you can use or adapt in a paper form or simple digital survey:
You can strengthen question design by comparing your findings later with department-level review work, such as the spring term AI audit scorecard for departments. Tutor-time evidence shows student patterns; departmental evidence shows where curriculum and assessment may need to respond.
The discussion around the survey matters as much as the survey itself. Introduce it calmly. Tell students the purpose is to help the school make sensible decisions based on reality, not rumours. Say clearly that you are not asking for names or confessions. You want trends, uncertainty and honest feedback.
It helps to script the opening. A tutor might say: “We know students are hearing mixed messages about AI. Before schools make rules, we want to understand how young people are actually using these tools, where the confusion is, and what guidance would help.” That framing lowers the temperature. It also makes it more likely that students will tell the truth about ordinary use.
Keep the discussion broad. Ask what students think their age group uses AI for, not who has done what. If a pupil starts naming others or describing a clear rule breach, redirect gently: “Let’s keep this general and anonymous.” The aim is to gather evidence, not create a disciplinary trail. This is especially important if your school is still resetting expectations after a break or policy refresh, as explored in the first week back tutor-time reset pack for AI boundaries.
Once responses are collected, resist the urge to focus on the most alarming comments. Start with frequency and clustering. Are students mainly using AI for revision support, drafting help, explanation, or shortcutting? Are younger pupils reporting confusion while older pupils report strategic use? Are there repeated mentions of pressure, panic, workload or late-night use?
Look, too, for contradictions. Students may say they know copying is wrong, yet also say they are unsure whether improving a draft with AI counts as cheating. That gap matters. It suggests the problem is not simply behaviour, but unclear instruction. If many students are using AI to revise, your next step may not be tighter sanctions. It may be better teaching on how to revise well with and without AI, perhaps linked to approaches like those in the mock season AI revision workflow.
You should also separate academic concerns from wellbeing concerns. If students mention using chatbots for reassurance, friendship or advice about difficult personal situations, that is a different category of response. It may require safeguarding messaging, digital literacy work and clearer routes to human support.
Ready to Revolutionise Your Teaching Experience?
Discover the power of Automated Education by joining out community of educators who are reclaiming their time whilst enriching their classrooms. With our intuitive platform, you can automate administrative tasks, personalise student learning, and engage with your class like never before.
Don’t let administrative tasks overshadow your passion for teaching. Sign up today and transform your educational environment with Automated Education.
The best response is proportionate. If the audit shows low-level, mixed and often uncertain use, a school-wide crackdown may be the wrong answer. Instead, you might produce clearer examples of allowed, limited and prohibited use. You might rewrite homework guidance so students know when AI can support planning but not final submission. You might ask departments to review tasks that are too easy to outsource, drawing on work such as the subject-by-subject AI-resilient assessment design guide.
Pastoral action may be just as important as academic action. If students are turning to AI because they feel stuck, the school may need better homework-help routines, revision scaffolds or signposting for emotional support. If they are using AI because they do not understand the point of an assignment, that is feedback on task design and communication.
A simple staff briefing can go a long way. Share three or four patterns, not a flood of raw comments. Then identify what changes now, what needs further review, and what remains non-negotiable. This keeps policy rooted in evidence rather than reaction.
After the audit, avoid using the results as proof that “students cannot be trusted”. That message will shut down honesty next time. Do not publish dramatic quotes out of context. Do not turn anonymous findings into a hunt for likely offenders. And do not assume all AI use is equally serious.
It is also unwise to respond with rules so broad that they cannot be enforced or explained. Students cope better with concrete examples than sweeping bans. If your school wants to build understanding rather than fear, it can help to pair policy language with current, balanced discussion of AI’s educational impact, such as in ChatGPT turns 3: education impact review.
A one-page repeatable template is enough. Keep the same ten questions, add the date and year group, and leave space for three summary notes from the tutor: common uses mentioned, common confusions, and any wellbeing themes requiring follow-up. Run it once a term with a sample of tutor groups or year cohorts. Over time, trends become more useful than one-off snapshots.
That repeated rhythm helps schools notice movement. Are students becoming more confident about boundaries? Are risky uses increasing? Are policy changes reducing confusion? Good AI governance is rarely about one big statement. More often, it comes from small cycles of listening, explaining and adjusting.
If schools want sensible rules, they need a clearer picture of student reality first. A 30-minute tutor-time audit is not complicated, but it can stop you from solving the wrong problem.
May your next policy meeting be guided by evidence, not guesswork.
The Automated Education Team
Microsoft Build 2026 brought a fresh wave of Copilot announcements, but …
Summer term is often the last realistic window for school leaders and AI …
Schools do not need a whole-school AI rollout to learn what works. In many …
Once the papers are marked, many departments want feedback that is sharper …
Primary assessment week can feel intense, especially when SATs and spring …
Exam week often creates pressure to do more, faster, with less time to …
Many school AI rules still rely on vague disclosures such as “I only used …
AI voice tools for modern foreign languages have improved sharply by 2026, …
Spring assessment season puts pressure on teachers, pupils and families to …
Enne kui koolid AI-reegleid karmistavad, aitab teada, mida õpilased tegelikult teevad. Lühike …
Innan skolor skärper reglerna för AI hjälper det att veta vad elever faktiskt gör. En kort …
Ennen kuin koulut kiristävät AI-sääntöjä, on hyödyllistä tietää, mitä opiskelijat oikeasti tekevät. …