UK universities struggle to catch AI cheating, detectors fail most of the time
Universities in the UK are caught between a rock and a hard place over generative AI cheating. Detection tools barely work. Studies show AI detectors catch less than 40% of AI-written text, dropping to just 22% in tricky cases where students try to hide their use of AI.
The problem? AI leaves almost no trace, so proving misconduct is nearly impossible unless students admit it. Some universities have cracked down hard, but many shift to “secure” in-person exams or design assessments assuming AI use. 
The idea that universities choose to ignore AI cheating just to protect revenue from international students doesn’t hold up, says Josh Freeman from the Higher Education Policy Institute.
“The last study he cites (Perkins et al, 2024) shows that AI detectors were accurate in fewer than 40% of cases, and that this fell to just 22% of ‘adversarial’ cases – when the use of AI was deliberately obscured. In other words, AI detectors failed to spot that AI had been used three‑quarters of the time.”
Josh Freeman, Policy manager, Higher Education Policy Institute
Turnitin’s AI checker, often promoted as a solution, is far from reliable. Research finds it missed 84% of AI-generated texts in tests. One paper notes detection tools tend to label AI text as human-written rather than detecting it properly.
“The researchers conclude that the available detection tools are neither accurate nor reliable and have a main bias towards classifying the output as human-written rather than detecting AI-generated text”
Prof Paul Johnson, University of Chester
Calls to return to traditional exams to curb AI use get pushback, too. University of Aberdeen’s Prof Robert McColl Millar says exams often fail to capture true writing skill and suggests analytical assessments with new material might work better.
“I would call for a move towards more analytical assessment, where students are faced with new material that must be considered in a brief period… this focus also helps students move towards application of new understanding, rather than a passive digestion of ideas.”
Prof Robert McColl Millar, Chair in linguistics and Scottish language, University of Aberdeen
The AI cheating debate isn’t settling anytime soon. For now, UK universities juggle unreliable tools, alternative assessments, and revenue concerns — with no easy fixes.
