Image credit: Author using ChatGPT
The conversation around Generative AI in education continues to grow at pace. Across the sector, institutions are grappling with both the opportunities and challenges presented by AI-powered tools. Much of the current discussion centres on academic integrity and how educators can adapt assessment practices in response to rapid technological change.
Current research often advocates moving away from a policing approach and towards fostering a culture of academic integrity. I strongly support this direction. Creating environments where students understand and value academic honesty is essential. However, a culture of integrity still needs robust academic misconduct detection processes. As Generative AI becomes increasingly sophisticated, institutions should combine educational and preventative approaches with effective methods for identifying and investigating potential misuse (Bittle and El-Gayar, 2025).
Traditionally, institutions have relied heavily on markers, reviewers and moderators to identify potential academic misconduct. However, emerging evidence suggests that academic expertise alone is no longer sufficient to reliably identify AI-generated or externally authored work. Fleckenstein et al. (2024), for example, found that both novice and experienced teachers struggled to distinguish AI-generated essays from student-written work, while Waltzer, Pilegard and Heyman (2024) similarly found that subject expertise did not significantly improve instructors' ability to identify AI-generated writing.
The challenge, therefore, should not simply be framed as a deficit in staff capability. Increasingly, it is also about whether staff have access to the appropriate evidence, technological tools and institutional processes needed to support informed judgement. Ajit, Maikkara and Ramku (2024) argue that detecting authorship-related misconduct is demanding and that decision-support tools can strengthen detection by bringing together multiple sources of evidence, while retaining human academic judgement at the centre of the process.
This is particularly important because no individual technological indicator should be treated as proof of misconduct. Rather, authorship technologies, writing-process evidence, learning analytics and student discussion can provide additional evidence that enables experienced staff to make better-informed and more consistent decisions.
One approach I have recently begun exploring is what I refer to as 100% screening of assessments. Where institutions have access to authorship or writing-process indicators, these can create opportunities to move from reactive investigation towards more proactive risk identification.
By screening all submissions and analysing available authorship indicators at scale, reviewers can identify patterns and potential red flags earlier. When combined with other quality assurance processes, 100% screening can help institutions focus investigative resources on the cases that genuinely require deeper review.
There is an important caveat here: authorship reports should never be treated as definitive proof of misconduct. They represent one strand of evidence among many. Context, assessment design, student explanations and additional supporting information must always be considered before reaching conclusions.
Nevertheless, stylometric technologies can provide markers and reviewers with additional information by analysing patterns in writing style and identifying possible changes of author within a document. Although the accuracy of these methods varies depending on the task, and baseline authentic writing submissions they can help highlight submissions that may require further human review (Albaqami et al., 2026).
As educational institutions continue to respond to the challenges posed by Generative AI, it is vital that we avoid seeing academic integrity and academic misconduct detection as competing priorities.
A strong culture of integrity should always be the goal. At the same time, robust detection processes and evidence-based technologies remain essential safeguards. The future is unlikely to be defined by either prevention or detection alone. Instead, success will come from combining both approaches in a balanced and proportionate way.
Authorship verification technology is not a silver bullet. However, when supported by mature processes and experienced investigators, it can become a powerful component of an institution's academic integrity framework.
The challenge now is not simply adopting new tools. It is ensuring that the foundations required to use them effectively are put in place. Only then can institutions use these tools effectively to help protect the value and credibility of education.
References
Ajit, S., Maikkara, A. and Ramku, W. (2024) ‘A decision support system to aid detection and substantiation of contract cheating in higher education’, in Chova, L.G., Martínez, C.G. and Lees, J. (eds.) EDULEARN24 Proceedings: 16th International Conference on Education and New Learning Technologies. Valencia: IATED, pp. 3503–3510. Available at: https://doi.org/10.21125/edulearn.2024.0911
Albaqami, H., Ayub, M.A., Ahmad, N., Ahmad, Y., Alqahtani, M.M., Algamdi, A.M., Owaidah, A.A. and Ahmad, K. (2026) ‘Stylometry analysis of human and machine text for academic integrity’, Computers, 15(4), article 217. Available at: https://doi.org/10.3390/computers15040217
Bittle, K. and El-Gayar, O. (2025) ‘Generative AI and academic integrity in higher education: A systematic review and research agenda’, Information, 16(4), article 296. Available at: https://doi.org/10.3390/info16040296
Fleckenstein, J., Meyer, J., Jansen, T., Keller, S.D., Köller, O. and Möller, J. (2024) ‘Do teachers spot AI? Evaluating the detectability of AI-generated texts among student essays’, Computers and Education: Artificial Intelligence, 6, article 100209. Available at: https://doi.org/10.1016/j.caeai.2024.100209
Waltzer, T., Pilegard, C. and Heyman, G.D. (2024) ‘Can you spot the bot? Identifying AI-generated writing in college essays’, International Journal for Educational Integrity, 20, article 11. Available at: https://doi.org/10.1007/s40979-024-00158-3
Annie Chohan is an Academic Integrity Lead at Global Banking School, with a specialist interest in academic integrity, authorship verification, generative AI and technology-enabled approaches to assessment assurance.
The authors' views are their own.
Thank you for being a member of ICAI. Not a member of ICAI yet? Check out the benefits of membership and consider joining us by visiting our membership page. Be part of something great!