Technology empowerment or risk avoidance? – investigating the overlooked factors influencing college students' acceptance of generative artificial intelligence

Generative artificial intelligence (GAI) is transforming higher education by enhancing learning tools, enabling personalized instruction, and streamlining administrative processes. However, colleges around the world have divergent policy positions on its educational use. In China, the use of international platforms such as ChatGPT is explicitly restricted within higher education institutions, while domestic colleges are simultaneously encouraged to explore effective pathways for integrating GAI into teaching and learning. Against this unique regulatory and educational background, this study examines Chinese college students' adoption of GAI using an extended Unified Theory of Acceptance and Use of Technology (UTAUT) model that incorporates perceived risk as an additional construct. Structural equation modeling (SEM) was applied to analyze survey data from 524 students. The findings reveal that performance expectancy and social influence significantly increase students' intention to use GAI. Facilitating conditions and behavioral intention also strongly predict actual usage behavior. Furthermore, gender and prior experience moderate the effects of social influence and performance expectancy on intention to use, while academic year has no significant moderating effect. This study identifies the key determinants and mechanisms shaping GAI adoption among Chinese college students under policy-driven constraints. The results offer empirical insights to inform the effective promotion and governance of GAI in education, both within China and in other countries undergoing technological regulation and digital transformation in the higher education sector.