Academic integrity has become a pedagogy of suspicion

Universities increasingly tell students that they are trusted partners in learning, yet institutional behaviour often suggests otherwise.

Over more than 25 years in higher education, I have come to believe that there are broadly three kinds of students who find themselves referred for academic misconduct. There are those who knowingly seek to deceive. There are those who could have done more to understand and comply with academic expectations. And there are those who genuinely try their best but struggle with conventions they are unfamiliar with. Unfortunately, our systems often struggle to distinguish between these groups.

When plagiarism detection software was first introduced, institutions often targeted students about whom they were already suspicious. Over time, that suspicion widened, and today we routinely scan all assessments through plagiarism detection software – at the same time as we monitor attendance, analyse engagement data and investigate the use of artificial intelligence (AI). Each of these practices may be justified in isolation, but their cumulative effect is unmistakable. Students are increasingly treated as risks to be managed rather than as learners to be supported.

This expansion of what we might call the pedagogy of suspicion is often justified in the moralised language of academic integrity. But much of what we classify as misconduct consists not of ethical failings but of unevenly distributed opportunities to acquire academic skills.

Some students arrive at university already familiar with what educational researchers call the “hidden curriculum”: accepted citation practice, understanding when collaboration becomes collusion, the ability to paraphrase appropriately. They benefit from intergenerational experience of UK higher education and schools with high progression rates into university.

But others do not. International students may have been taught under different academic conventions, and students with specific learning needs may rely on technologies increasingly powered by AI.

In many institutions, the boundaries of acceptable AI use remain unclear. Yet our response to the rise of the technology has often been to search for linguistic differences that might indicate machine authorship, inevitably placing greater suspicion on non-native English speakers and some neurodiverse students, whose writing naturally varies in ways that automated systems and human markers can misinterpret as “inconsistent”.

I am not, of course, arguing that deliberate cheating does not exist. A small number of students knowingly seek to deceive, just as they always have. They deserve processes that are fair, robust and proportionate. Upholding standards still matters as much as it ever did.

But by collapsing intentional deception and imperfect participation in academic conventions into the same category, we have created misguided, quasi-judicial processes that add stress and anxiety to already stressful assessment systems.

We could achieve much of what current academic integrity systems seek to accomplish by treating academic practices as competencies to be taught, developed and assessed transparently. Expectations around citation, paraphrasing and appropriate collaboration could be embedded explicitly within learning outcomes and marking criteria, just as we assess other disciplinary skills. If referencing matters, assess it. If source integration matters, assess it. If responsible AI use matters, assess it. Most importantly, teach these practices not as optional extras, but as explicit and developmental parts of the curriculum.

Students who fail to demonstrate those competencies would still fail modules, but we would move away from the assumption that every error requires an investigation, and every misunderstanding represents a failure of moral character.

For the relatively small number of students who deliberately seek to deceive, institutions need systems that are genuinely fit for purpose. At present, our ability to detect sophisticated and often expensive forms of cheating remains uneven. Students with highly educated parents who can proofread their work are more likely to avoid suspicion than those who use an open-access AI model to correct their work. So are students who can afford the bespoke services of essay mills – and, indeed, those who pay for subscription AI services. Until we can address that problem, we should be wary of subjecting some groups of students to greater scrutiny than others.

Universities rightly speak of belonging, partnership and student engagement. Yet these aspirations sit uneasily alongside systems built upon suspicion. Belonging is difficult to foster when trust is conditional.

The question facing higher education, then, is not simply how we maintain standards, it is whether we want universities organised around trust or suspicion. In my view, the latter does not offer an answer to the challenges posed by AI and academic misconduct. What we need, instead, is better, clearer and more equitable education.

Nick Cartwright is an associate professor and director of student success in the School of Law at the University of Leeds.