Perceived artificial intelligence dependence and anxiety symptoms among university students: a three-wave study of experiential avoidance and digital literacy

University students increasingly use artificial intelligence tools for learning, decision-making, and academic work, making perceived AI dependence a relevant issue for public mental health and digital public health rather than only for educational technology. This three-wave time-lagged survey examined whether experiential avoidance functioned as a putative explanatory process linking perceived AI dependence with later anxiety symptoms and whether digital literacy moderated the tested pathways. Online survey data were collected from 1,451 university students with AI-use experience; perceived AI dependence, digital literacy, and baseline demographic variables were measured in January 2026, experiential avoidance was measured in March 2026, and anxiety symptoms were measured in June 2026. PLS-SEM results showed that experiential avoidance was strongly and positively associated with later anxiety symptoms (beta = 0.677, p < 0.001), whereas the direct association between perceived AI dependence and anxiety symptoms was not significant (beta = -0.030, p = 0.147); perceived AI dependence was negatively associated with experiential avoidance (beta = -0.114, p < 0.001), producing a statistically significant but theoretically unexpected negative time-ordered indirect association (beta = -0.077, p < 0.001), and digital literacy did not significantly moderate the tested pathways. The findings suggest that anxiety symptoms in AI-mediated learning contexts are closely linked to avoidance-related psychological processes, while the meaning of perceived AI dependence is context-dependent and should not be equated automatically with harmful use or clinical addiction; campus mental-health promotion should therefore integrate responsible AI education with support for psychological flexibility, uncertainty tolerance, and autonomous learning.