The new AI in Teaching and Learning Policy in practice

18 Sep 2026

An overview from Mark Carrigan, our Associate Dean for Digital Education, of what teaching colleagues need to do now in line with the University’s new AI in Teaching and Learning Policy.

We’re about to begin teaching, which means conversations with students about AI will be taking place soon. This is a reminder of what the new AI in Teaching and Learning Policy means in practice and what you need to do this academic year.

If you are a unit lead, please make sure you have chosen one of the four categories for each summative assessment. The intention is to make sure our students have a clear statement of how we expect them to use or not use AI in their assessed work. It does not mean changing what you do or what you expect from students, only making that practice and those expectations explicit in a way that will be consistent across their units.

1. AI Prohibited means that students should not make any use of AI in the assignment
2. AI Minimal means that students should only use AI in limited ways that contribute to the presentation without changing the ideas, arguments or conclusions
3. AI Permitted means that students can use AI in specific ways which are communicated to them but not beyond these
4. AI Integrated means that the assessment is designed in a way where the use of AI is necessary in order to complete it.

If you are unsure about which category to use, an online tool helps distinguish between different types of use which students might make of AI in your unit. In my last AI policy update on StaffNet, I suggested some questions which might be useful to think about when making this choice. There is also a videocast of FSE colleague Rachel Parker-Strak talking through her own experience of applying the policy.

Once you have chosen the category, it is important to make this visible to students through Canvas and the assignment brief, along with an explanation of what this means for your unit. A short screencast on Canvas, talking students through the assessment and why you've chosen its category, is one of the most effective ways of doing this and it takes far less time than it sounds.

What matters is that you communicate your expectations to students. You know your students and we trust your judgement about what’s right for them. The intention of the policy is to ensure that we are having these conversations in a consistent way across the Faculty. This means we’d like you to talk about the category with your students. Why have you chosen this category? How do your expectations relate to the learning objectives for the unit? What are the specific risks you see for student learning if they misuse AI for the assignment?

There is guidance on how students can use AI, and what to tell them, on the Copilot SharePoint site, including slides which can be adapted and incorporated into lectures, workshops and seminars. Try to make space for discussion where you can and ensure students have the opportunity to ask questions and raise issues about the use of AI. If students ask what happens if they exceed the category, there are two answers:

1. If their assignment presents AI-generated content as if it were their own, this counts as plagiarism under existing malpractice procedures, whatever the category.
2. If students use AI in an assessment category where it is not permitted, their work will often fail to demonstrate the skills and learning outcomes being assessed and should be marked accordingly. For example, an assessment designed to develop original analysis may receive a lower mark if the contribution of the AI limits the student's demonstration of critical thinking.

The categories describe what each assessment is designed to develop in students. They are not rules with separate penalties attached.

This won't resolve the challenges to assessment integrity. I'd rather be direct about the limits. The policy tells students what we expect of them. It was never designed to enforce that, because a categorisation scheme can't.

The work that will actually secure assessment, through secured exams, stress testing the assessments most at risk, and supporting redesign, is under way at University level. We'll share details as they're confirmed, and I'd expect Semester 2 to be the realistic horizon for most of it.

Formal guidance on responding to suspected breaches will follow early in Semester 1, informed by consultation with teaching and learning leadership across the four Faculty Schools. In the meantime, issues are being collected through a fortnightly meeting with School leads, so if something isn't working, please tell your school digital learning lead or get in touch with me.

Dr Mark Carrigan

Associate Dean for Digital Education, Faculty of Humanities