Discussion

Edinburgh University recorded 245 AI-related misconduct cases in 2024/25

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13715

The University of Edinburgh recorded 245 student misconduct cases involving AI in 2024/25, up from 78 the year before, according to figures obtained through freedom of information requests. The figures show a sharp rise, but the university cautions that changes in how cases were categorised make the early totals incomplete.

Watch Desk analysis

What happened

The figures come from two FoI responses obtained by The Herald and reported on 14 November. Resultsense reports that the university recorded 19 AI-related cases in 2022/23, 78 in 2023/24 and 245 in 2024/25. The 2024/25 total was just under a third of the university’s 795 proven misconduct cases that year.

The university says schools did not use a separate AI category consistently until 2023/24. Some earlier cases may have been recorded under cheating or plagiarism instead, so the increase may partly reflect improved recording. Figures for 2025/26 are expected around February 2027.

The university’s guidance lists actions that count as misconduct, including submitting AI-generated text, code or other material without declaring it, using an AI translator on assessed work, and citing AI-suggested sources without checking them. It also prohibits letting an AI agent or AI-powered browser carry out tasks in the university’s online learning platform.

Why it matters

These numbers make the university’s AI rules more than a theoretical policy question: in 2024/25, AI-related cases accounted for nearly one in three of its proven misconduct cases. But the inconsistent early classification means the trend should not be read as a clean measure of how much student AI misuse has grown.

The rules also reach beyond chatbot-written essays. They address tools that can act within online learning systems, putting the line between assistance and having software do assessed work squarely in view.

Our read

The case count is striking; the caveat is essential. Better recording could explain some of the jump, but the 2024/25 figures still show that AI is now a substantial part of the university’s recorded misconduct workload. And a policy covering agents as well as generated text is a sign that universities are having to define acceptable use for tools that do more than draft.

What to watch

  • The university’s 2025/26 figures, expected around February 2027.
  • Whether future reporting makes AI-related categories consistent across schools.
  • How universities explain and enforce rules on AI agents and browsers in learning platforms.

Discussion spark: Should universities treat an AI agent completing tasks in a learning platform as misconduct by default, or should the rules depend on what the tool did and whether the student disclosed it?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.