AI content detection in student text assignments, false positive rates, and academic integrity

Universities increasingly use AI content detection tools to identify student assignments potentially produced by generative AI, yet concerns remain regarding their reliability and false-positive classification of human-written work. This systematic review examines the causes, consequences, and mitigation of false-positive AI detection in academic contexts. The review covers literature published between 2020 and 2026 and includes 90 eligible studies, comprising 81 Scopus-indexed publications and nine studies identified through Google Scholar. Bibliometric mapping using VOSviewer was combined with Structural Topic Modelling to examine keyword relationships, latent research themes, and temporal development, while qualitative synthesis identified major causes, impacts, and mitigation strategies. The findings reveal three broader themes concerning academic integrity and student assessment, plagiarism and responsible AI governance, and detection accuracy, reliability, and false positives. False-positive outcomes vary substantially across tools and evaluation settings, limiting direct quantitative comparison. Major contributing factors include algorithmic and language bias, limited contextual understanding, surface-level linguistic features, and insufficiently diverse training data. The evidence further highlights risks to assessment fairness, student confidence, and institutional decision-making. Effective responses therefore require transparent detection practices, human review, AI literacy, clear institutional policies, and accessible appeal procedures.