Thousands of university students across the UK are cheating using AI, according to Freedom of Information requests from hundreds of UK universities accessed by The Guardian.

A report from the news outlet found that as AI tools, particularly ChatGPT, becomes more mainstream, more and more students are turning to them for academic misconduct.

The Guardian issued FOI requests to 155 universities to see data on proven cases of academic misconduct, ranging from plagiarism to AI misconduct, over the past five years.

Interestingly, cases of plagiarism have fallen over the past five years as AI takes hold. In 2019-2020, plagiarism made up almost two-thirds of all academic misconduct cases.

As AI became more widely available, plagiarism fell from 19 per 1,000 students to just 15.2 in 2023-24, and is expected to fall further in the coming year.

The research however found nearly 7,000 proven cases of academic misconduct using AI in 2023-24, which is about 5.1 per 1,000 students. In 2022-23, this figure sat at 1.6 per 1,000.

It is expected that these figures will rise this year, as plagiarism cases continue to decrease.

Though 155 universities were contacted, only 131 provided data, and out of these, 27% had yet to record AI misuse as a separate academic misconduct category in 2023-24.

Recommended reading

• 84% of UK Students Use AI—But Are They Being Sufficiently Supported?
• Scots Computing Science at Crisis Point as Teacher Numbers Crash
• 80% of Teachers Believe Students Cheat Essays By Using AI

AI cheating or misuse is becoming a more common issue, with many academic institutions ill-equipped to deal with the surge.

Research from the University of Reading found that despite systems in place for assessments, students were often able to submit AI-generated work without it being detected 94% of the time.

Resources are circulating on social media helping students to use AI for school assignments without getting caught, touting prompt suggestions or AI tools to create more ‘human’ outputs that can go undetected.

Further, AI misuse can be difficult to discern – the outright use of AI is not prohibited by most institutions, and can be used to brainstorm, generate structures, or provide suggestions. Crossing the line from use to misuse can be difficult for institutions to lay out in AI guidance, and can be even more difficult to police.