Is Writing Too Well a Crime? St Andrews Medical Student Wrongly Accused of Using AI

In August last year, medical student Harrison Sharples received a warning from his university accusing him of 'academic misconduct' — specifically, that the essay he had written over the summer had been flagged as artificially generated by artificial intelligence (AI). He said he was 'shocked' by the accusation.

The 20-year-old student at the University of St Andrews in Scotland could not understand how his marker might believe the work wasn't his own. He told BBC that he never even considered using AI because he had ample time to complete the assignment and genuinely enjoyed the writing process. 'To be honest, I often forget that this is even an option available to me,' he emphasized to reporters.

According to the university, his paper examined the impact of play therapy on reducing anxiety in hospitalized children and was assessed by two human graders. One grader expressed concern about possible AI use.

Grading notes seen by BBC raised concerns about the work’s 'polished/uniform tone.' The reviewer also noted 'consistently flawless grammar, similar sentence structures, and no variation in style.' Annotations mentioned a 'formulaic paragraph structure following a typical AI pattern: general claim, example/result, and conclusion.'

Other observations pointed to repeated phrasing, lack of 'nuanced criticism,' and extensive use of em-dashes — a punctuation mark believed to be favored by AI models.

Due to these concerns, Harrison was scheduled for an online hearing hosted by a panel of experts from the university's medical school to defend himself. He said he had prepared thoroughly with evidence and arguments, but on the day of the hearing, the virtual meeting link was not delivered in time. After emailing the university, he only received the link five minutes after the hearing had started.

'At that moment, I was extremely nervous because I felt this was my one chance to prove myself and clear my name — and they might think I simply didn’t show up!' he said.

During the half-hour hearing, the student was asked to define medical terms used in his essay, which Harrison said surprised him. 'I was really shocked when the hearing began,' he said. 'I don’t feel I presented myself at my best.'

A week later, he received a ruling confirming the AI-use allegation. This meant his grade was capped at 7 — just passing — and he might have to declare this incident when applying for provisional registration with the UK’s General Medical Council (GMC). 'That’s exactly why I decided to appeal,' Harrison said.

The university accepted the appeal, acknowledging that the delayed delivery of the link affected his ability to present his case. He was invited to another hearing in mid-November.

Harrison said the second session was similar to the first, but this time the panel focused on his writing process. 'It went much more smoothly,' he said. 'I felt I communicated my points far more clearly.'

This time, the university decided not to uphold the allegations against him.

'I felt a huge sense of relief,' Harrison said. 'It would’ve been such a burden if this violation stayed on my record.' His paper ultimately received full marks, and the incident will not affect his professional record.

However, Harrison said the entire episode weighed heavily on him throughout the semester, affecting his studies. He told reporters that although he had support from friends, his girlfriend, and family back home in Manchester, the experience still impacted him 'deep down.'

'There were so many times I just didn’t know what was happening, and I think that’s what worried me most and caused the most stress,' he said. 'This matter occupied too much space in my mind, leaving no room to think about anything else. That’s why I neglected my studies — my brain was completely consumed by this one thing.'

As a result, Harrison had to postpone his exams by a semester, now taking them in mid-summer. His graduation ceremony has also been delayed until December this year. 'If it weren’t for what happened to me, I would’ve graduated already,' he said.

AI ethics consultant Sidrah Hassan told BBC that the pace of development in large language models (LLMs) is 'incredibly fast,' making it difficult for institutions to keep up. 'AI systems often leave telltale signs in their outputs that help identify whether something was AI-generated or human-written,' she said.

But she added that existing AI detection tools are not robust enough and have high error rates. While human markers can flag uncertain cases, Hassan stressed that penalizing students may not be sustainable. 'I don’t think that’s the point,' she said. 'We should change our testing methods because, as humans, we’ll always look for the easiest way to do things.'

In response, the University of St Andrews issued a statement: 'We do not comment on individual case details, but we can confirm that in this instance, the university supported the student’s appeal, which related solely to procedural issues concerning the hearing arrangements.'

'We are working within the university and across the sector to support responsible use of AI in academic work, including in medicine.'

This article was originally written in English. We used AI to assist with translation, and it was reviewed by BBC journalists before publication. Learn more about how we use AI.