Preparing for the new AI in Teaching and Learning Policy

25 Aug 2026

An update from Mark Carrigan, Associate Dean for Digital Education, Faculty of Humanities

Our new AI in Teaching and Learning Policy takes effect from 1 September 2026. This means there are preparations we need to make in the Faculty of Humanities by 22 September 2026.

The policy introduces a common framework for communicating expectations to our students of how AI can be used in assessed work - and sets out a clear, proportionate ask of teaching colleagues ahead of the new academic year.

The new policy asks unit leads to articulate their expectations around the use of AI by students and to ensure these are communicated in a clear and effective manner. This does not mean changing existing assessment practice but rather ensuring that what you ask of students in relation to AI-use is made entirely clear to them.

In practice this means there are two things which each unit lead needs to do by 22 September 2026:

1. Classifying each of your assessments in one of the four categories

The four categories describe how far students may use AI in an assessment, ranging from:

A.    complete prohibition

B.    use for presentation and polishing

C.    more expansive forms of specified use

D.    to AI being required to complete the assessment.

Here are some questions you might think about when classifying your assessments. They are prompts rather than something to complete, and if you are teaching a unit for the first time you might read the first as a question about what you would expect:

1. How do you believe students have used AI on this assessment previously?
2. What did you notice that led you to form these beliefs? How reliable do you think those signals are?
3. Which of those uses are you comfortable with? Can you articulate why?
4. Which of those uses do you find problematic? Can you articulate why?
5. Did you talk to students about their AI use? Did students ask you questions about it?

It can be helpful to discuss these within your teaching teams or programme teams. If this is not possible in time, you might discuss it with one colleague instead to learn from each other’s experience of the process. The hope is that we might deepen our understanding of our existing practice through this process.

Nothing here requires changes to your intended learning outcomes or assessments, and none should be made for September. If this process makes you want to change something, that is an important outcome and there will be more information provided about how best to do this.

There is clearly a degree of ambiguity between category B and category C. If you're struggling to place an assessment, the rule of thumb I would suggest is to ask whether AI use would change what the student is saying or only how they are saying it. Category B encompasses the latter, roughly equating to the help a proofreader might offer, where the ideas, the argument and the conclusions remain the student's own. Anything which extends into the substance of the assessment belongs in category C, where the requirement is to specify what is acceptable rather than leaving it open.

At this stage we expect only a handful of category D (AI-Integrated) across the Faculty, because it is the one category where students cannot choose to opt out. Please contact your school Director of Teaching and Learning by 30 August 2026 to discuss units in category D.

2. Communicating classification to students with appropriate guidance

These classifications need to be made available to students through Canvas on the relevant assessment page.

There are two interactive resources which can help you produce guidance for students about the different types of AI activity which are acceptable: available here.

These tools will not tell you which category your assessment belongs to because we trust you to make this judgement. The policy simply ensures that all colleagues make these judgements transparent to students in a clear and standardised way. There is a videocast of an FSE colleague talking through her own experience of applying the policy.

For both category B and category C, please ensure that students who prefer not to use these tools are still able to complete the assessment. If you are choosing category C, the assessment brief should explain how.

That is the baseline. Beyond it, the classification is a good starting point for conversations with students about what AI means for their learning in your unit, whether in class, in a Canvas screencast or through supplementary guidance. The key thing is that we make our expectations clear and begin those conversations.

The policy is a starting point rather than a solution. It won't solve the problems which AI creates for assessment, but if we can all reflect on our expectations then communicate them to students, we are moving in the right direction.

The new policy will iterate each year and all feedback will be gratefully received. There will be further support offered at school level in September and forums for professional dialogue about the policy, as well as AI use more generally, across Humanities throughout the next year.

If you have any further questions, please contact your School Director of Teaching and Learning in the first instance.