The Influence of AI Literacy on Academic Integrity in Higher Education

AI literacy is increasingly relevant to academic integrity, yet familiarity with artificial intelligence does not necessarily entail responsible academic conduct. This article examines how affective, behavioural, cognitive, and ethical dimensions of AI literacy relate to self-reported academic dishonesty, while considering students' beliefs about AI-generated knowledge. An exploratory secondary analysis of 972 Chinese student records is combined with a purposive literature synthesis and historical Ugandan tertiary-education indicators. Unweighted composite scores, reliability estimates, correlations, and ordinary least squares models with HC3 standard errors were calculated. Recorded demographic and institutional covariates accounted for 10.1% of outcome variance; adding the four literacy dimensions increased this to 16.6%, and adding epistemic-belief dimensions increased it to 19.4%. Ethical literacy was negatively associated with dishonesty in the full model (β = −.325, 95% CI [−.412, −.238]), as was justification (β = −.216, 95% CI [−.310, −.122]). Cognitive literacy had a small negative bivariate correlation but a positive adjusted coefficient (β = .143), cautioning against treating technical confidence as a sufficient indicator of integrity. Repeated response profiles, correlated scales, self-report, and unresolved coding information limit inference. The article proposes an integrity-oriented assessment sequence question, trace, test, and own that connects literacy instruction with inspectable evidence, attribution, and student responsibility. The findings support a differentiated account of AI literacy, not a causal claim or an estimate of Ugandan student behaviour.