<p>With the growing adoption of generative artificial intelligence (GenAI) by higher education students across various learning contexts, its associated ethical risks have garnered considerable attention. However, a validated scale for measuring higher education students’ perceptions of GenAI ethical risks remains unavailable. To address this gap, this study systematically investigates the multidimensional structure of GenAI ethical risks through scale development and validation. An exploratory factor analysis (EFA) was conducted on 290 students to establish the scale structure, followed by a confirmatory factor analysis (CFA) on data from an additional 676 students to validate the structure and evaluate competing models. The results identified four dimensions: subjective ethical risk, algorithmic ethical risk, relational ethical risk, and ecological ethical risk. The final scale consists of 20 items and demonstrates satisfactory psychometric properties, including robust reliability (α = 0.934), as well as convergent, discriminant, and criterion validity. Furthermore, the second-order factor structure was empirically supported, indicating the scale’s theoretical coherence. Beyond its theoretical contributions to understanding higher education students’ perceptions of GenAI ethical risks, this scale offers practical value by equipping educators with an evidence-based tool for systematic risk assessment in educational settings. It enables the development and implementation of informed policies governing the ethical use of GenAI in higher education.</p>