Low student self-efficacy is concerning. The mediating role of AI- supported self-regulation learning on the relationship between peer pressure and perceived student self-efficacy in mathematics was verified. Predictive research design was used. The data were from 268 selected Grade 11 senior high school students, selected via stratified random sampling. They were analyzed using mediation analysis. Findings showed that the determinant-criterion correlation was partially but significantly mediated, partly supporting Social Cognitive Theory. Future research may explore other mediating variables to explain the remaining 29% variance, while school leaders may enhance student perceived self-efficacy through AI-supported self-regulation tools, teacher training, and positive peer learning environments.