Abstract
<title>Abstract</title> <p>The rapid adoption of Generative Artificial Intelligence (GenAI), particularly ChatGPT, has transformed higher education by offering novel pedagogical affordances while simultaneously challenging traditional frameworks of academic integrity. This study presents a systematic review and meta-analysis of 53 empirical studies, conducted according to PRISMA guidelines, to identify determinants of GenAI misuse and dependence among university students. The studies were retrieved from Scopus, Web of Science, Dialnet, and PubMed, and further complemented with a manual search. The systematic review indicates that GenAI misuse is consistently associated with Dark Triad personality traits—specifically Machiavellianism, psychopathy, and narcissism—as well as moral disengagement and younger age. These traits are associated with procrastination in an academic context that has ambiguous cheating policies alongside high workloads and time pressure. On the other hand, ethical awareness and perceived consequences function as protective factors against cheating and plagiarism. Meta-analytic results revealed that the association between these variables ranged from low to moderate. In terms of GenAI dependence, the findings reveal a strong link to psychological factors such as high neuroticism, depressive symptoms, and anxiety together with low self-efficacy and an overestimation of GenAI to manage academic tasks. All these factors facilitate procrastination, establishing a compulsive use of GenAI which is linked to low academic performance and reduced critical thinking skills. The meta-analysis findings highlight that GenAI dependence is inversely associated with the above-mentioned psychological factors, academic performance, and critical thinking abilities, with most values being moderate. These findings provide a coherent framework for developing evidence-based interventions to foster a responsible integration of GenAI in educational settings.</p>