When AI Anxiety Coexists With Capability: ICT Self‐Efficacy and Institutional Support in Preservice Teacher Education

ABSTRACT Preparing preservice teachers for AI‐enabled teaching requires understanding anxiety alongside perceived capability and self‐regulated learning (SRL). Drawing on data from 12,147 third‐year preservice teachers across 17 Chinese institutions, this study combined structural equation modelling, observed‐score moderation, and latent profile analysis. Four profiles emerged: the Moderate Profile, the Lower‐Efficacy, Lower‐SRL Profile, the Low‐Anxiety, Higher‐SRL Profile, and the High‐Anxiety, Higher‐SRL Profile. The two higher‐SRL profiles shared high ICT self‐efficacy despite contrasting anxiety levels. AI anxiety was positively associated with ICT self‐efficacy, which was strongly associated with SRL. Perceived institutional ICT support strengthened the anxiety–efficacy association. These complementary findings identify perceived capability as a key differentiator of regulated learning across anxiety levels. Teacher‐education programmes can assess anxiety, efficacy, and SRL together and provide practice opportunities and institutional resources that support learning.