When ChatGPT becomes a shortcut: psychological predictors of student plagiarism threshold

Generative Artificial Intelligence (GAI) tools such as ChatGPT are increasingly integrated into higher education and challenge traditional norms of academic integrity. However, we lack empirical evidence on the psychological mechanisms linking students’ perceptions to plagiarism decisions. The objective of the present study was to investigate how students’ perceptions of ChatGPT’s Competence predict Plagiarism Threshold, defined as the extent to which submitting ChatGPT-generated assignments is acceptable. A cross-sectional study was conducted with 235 Computer Science students. Path analysis was used to examine the indirect effects of Perceived Competence of ChatGPT on the Plagiarism Threshold. Higher Perceived Competence of ChatGPT was linked to higher Task Efficiency, which in turn predicted higher levels of Moral Disengagement and Plagiarism Threshold (β = .056, p < .01). In contrast, higher Perceived Competence of ChatGPT also predicted higher levels of Academic Engagement, which in turn predicted lower levels of Moral Disengagement, and Plagiarism Threshold (β = −.022, p < .05). EStronger perceived competence of ChatGPT does not always result in higher levels of Plagiarism Threshold. Instead, its role depends on the psychological context in which it is used. For some students, ChatGPT may serve as a useful tool that enhances learning and engagement, while for others, particularly under pressure, it may increase the Plagiarism Threshold, rationalising unethical behaviours as more permissible. More studies regarding student-ChatGPT interaction are essential for shaping more responsible educational policies that promote ethical use of GAI tools in academic settings.