Generative AI and Academic Integrity in Higher Education: Challenges and Future Directions
TL;DR
It is argued that detection-centred enforcement is a structurally weak control and proposed instead a layered institutional framework in which policy and governance, pedagogy and assessment redesign, and technology-based assurance operate as mutually reinforcing controls, sustained by a continuous audit and improvement cycle.
Abstract
The public release of ChatGPT in late 2022, and the wave of generative artificial intelligence (GenAI) tools that followed it, has unsettled long-standing assumptions about how learning is demonstrated and assessed in higher education. Essays, reports, code, and even reflective writing can now be produced in seconds by systems whose output is fluent, personalised, and largely indistinguishable from student work. This paper examines the resulting collision between GenAI and academic integrity from a governance and assurance perspective. Drawing on the rapidly growing literature published since 2022, it maps the principal challenges facing institutions: the unreliability and demonstrated bias of AI-text detection tools, the erosion of assessment validity, widening equity gaps, fragmented and reactive policy, limited faculty capacity, and the contamination of scholarly work by fabricated references and hallucinated content. The paper argues that detection-centred enforcement is a structurally weak control and proposes instead a layered institutional framework in which policy and governance, pedagogy and assessment redesign, and technology-based assurance operate as mutually reinforcing controls, sustained by a continuous audit and improvement cycle. Future directions are discussed, including two-lane assessment models, AI literacy as a graduate attribute, provenance and watermarking infrastructure, and the emergence of academic integrity as an auditable domain of institutional risk management