Cuproptosis is a novel form of cell death, mediated by protein lipid acylation and highly associated with mitochondrial metabolism, which is regulated in the cell. Ulcerative colitis (UC) is a chronic inflammatory bowel disease that recurs frequently, and its incidence is increasing worldwide every year. Currently, a growing number of studies have shown that cuproptosis-related genes (CRGs) play a crucial role in the development and progression of a variety of tumors. However, the regulatory role of CRGs in UC has not been fully elucidated. Firstly, we identified differentially expressed genes in UC, Likewise, CRGs expression profiles and immunological profiles were evaluated. Using 75 UC samples, we typed UC based on the expression profiles of CRGs, followed by correlative immune cell infiltration analysis. Using the weighted gene co-expression network analysis (WGCNA) methodology, the cluster's differentially expressed genes (DEGs) were produced. Then, the performances of extreme gradient boosting models (XGB), support vector machine models (SVM), random forest models (RF), and generalized linear models (GLM) were constructed and predicted. Finally, the effectiveness of the best machine learning model was evaluated using five external datasets, receiver operating characteristic curve (ROC), the area under the curve of ROC (AUC), a calibration curve, a nomogram, and a decision curve analysis (DCA). A total of 13 CRGs were identified as significantly different in UC and control samples. Two subtypes were identified in UC based on CRGs expression profiles. Immune cell infiltration analysis of subtypes showed significant differences between immune cells of different subtypes. WGCNA results showed a total of 8 modules with significant differences between subtypes, with the turquoise module being the most specific. The machine learning results showed satisfactory performance of the XGB model (AUC = 0.981). Finally, the construction of the final 5-gene-based XGB model, validated by the calibration curve, nomogram, decision curve analysis, and five external datasets (GSE11223: AUC = 0.987; GSE38713: AUC = 0.815; GSE53306: AUC = 0.946; GSE94648: AUC = 0.809; GSE87466: AUC = 0.981), also proved to predict subtypes of UC with accuracy. Our research presents a trustworthy model that can predict the likelihood of developing UC and methodically outlines the complex relationship between CRGs and UC.