Thousands of university students in the UK have been caught misusing AI tools like ChatGPT, with almost 7,000 proven cases in 2023-24, according to a Guardian investigation. This equates to 5.1 cases per 1,000 students, up from 1.6 per 1,000 the previous year. Early figures for the current academic year suggest a further rise to 7.5 per 1,000.
Experts warn that recorded cases represent only the tip of the iceberg, as many instances go undetected. A study by the University of Reading found that AI-generated work could be submitted without detection 94% of the time. Dr Peter Scarfe, co-author of the study, noted that proving AI misuse is far harder than traditional plagiarism, making it difficult to take action without falsely accusing students.
Traditional plagiarism has declined markedly, from 19 cases per 1,000 students in 2019-20 to a projected 8.5 per 1,000 this year. The shift reflects the growing sophistication and accessibility of generative AI tools. However, over 27% of universities did not yet record AI misuse as a separate category in 2023-24, indicating the sector is still adapting.
Students are using AI for brainstorming and structuring assignments, with many relying on paraphrasing tools to bypass detection. Dr Thomas Lancaster of Imperial College London said that when used skilfully, AI misuse is very hard to prove. One student, Harvey, a final-year business management student, said most of his peers use AI to generate ideas and references, but rarely copy text directly.
The findings highlight the challenge for universities in updating assessment methods. While some advocate for more in-person exams, Dr Scarfe acknowledged that moving all assessments offline is unfeasible. The sector must find ways to ensure learning integrity while embracing the reality that students will use AI, whether permitted or not.
