<![CDATA[The rapid adoption of generative artificial intelligence in higher education has intensified debates about whether ChatGPT supports responsible learning or weakens academic integrity through over-reliance and misconduct. This study examined the direct effects of ChatGPT usage and AI over-reliance on student integrity and tested academic misconduct as a moderating variable. A quantitative cross-sectional survey was conducted with 571 Indonesian undergraduate students who had experience using ChatGPT or comparable AI tools for academic purposes. Data were collected using a validated 20-item questionnaire measured on a five-point Likert scale and analyzed through Partial Least Squares Structural Equation Modeling (PLS-SEM) with 5,000 bootstrap resamples. The results show that purposive ChatGPT usage positively predicts student integrity (β = 0.482, t = 10.419, p < .001), whereas AI over-reliance has a significant negative effect (β = -0.215, t = 3.853, p < .001). Academic misconduct also shows a significant direct path to integrity (β = 0.159, p = .005) and moderates the relationship between AI over-reliance and student integrity (β = 0.247, t = 5.143, p < .001), but it does not significantly moderate the ChatGPT usage-integrity relationship. These findings demonstrate that the ethical consequences of generative AI depend on how students engage with the technology. The study implies that universities should not merely prohibit AI use, but should develop clear AI governance, digital ethics instruction, and assessment designs that encourage critical, transparent, and accountable use of ChatGPT.]]>
Copyrights © 2026
