TL;DR
The University of Edinburgh penalised at least 342 students for academic misconduct involving AI across 2022/23 to 2024/25, according to two FoI responses obtained by The Herald and reported on 14 November. Cases climbed from 19 in the first year the university recorded them to 78 in 2023/24 and 245 in 2024/25. Figures for 2025/26 are not expected until about February 2027.
The numbers
In 2024/25 the university recorded 795 proven misconduct cases in total, so AI-linked cases made up just under a third. The Herald calculates the 245 figure as a 1,190% rise on 2022/23 and a 214% rise on the previous year.
The university’s FoI response comes with a caveat. A separate AI sub-category was not used consistently across schools until 2023/24, so the early figures may be partial and some AI cases may have been filed as cheating or plagiarism instead. In our reading, part of the increase may therefore reflect better recording rather than more misuse.
The university’s student body exceeds 49,000. The share involved in misconduct cases of any kind edged up from 1.6% in 2023/24 to 1.65% in 2024/25.
What counts as misuse
University guidance lists the uses of generative AI that amount to misconduct. They include passing off AI output as original work, running assessed work through an AI translator before submission, and submitting AI-generated text, images, audio, video, formulae or code without saying so. The list also covers citing sources an AI turned up without reading and verifying them, and letting an AI agent or AI-powered browser do any task inside the university’s online learning platform.
Miles Briggs, a Scottish Conservative MSP for Edinburgh and Lothians East, said it was “very worrying” that hundreds of students had been willing to use AI to get ahead unfairly and called on the Scottish government to say how it will help universities deal with cheating.
Looking Forward
The figures echo concerns south of the border, set out in the Office for Students’ warning we covered in our report on AI and degree credibility. In our view, the guidance’s inclusion of AI agents and AI browsers is notable: Edinburgh is already writing rules for tools that can complete coursework on a student’s behalf, not just draft text. How institutions detect that kind of use will matter as much as how they define it.