This study analyzes the performance of ensemble learning algorithms in detecting phishing URLs using the PhiUSIIL Phishing URL dataset. The three algorithms compared are CatBoost, XGBoost, and LightGBM. The research stages include data preprocessing, data division into an 80:20 train-test split, and performance evaluation based on accuracy, precision, recall, and F1-score metrics. The results show that XGBoost has the best performance with an accuracy of 97.54% and an ROC AUC of 93.05%, followed by CatBoost with an accuracy of 97.46% and an ROC AUC of 92.94%. LightGBM, although it has lower performance, still shows good results with an accuracy of 96.99% and an ROC AUC of 91.85%. The data cleaning process successfully improves efficiency by eliminating irrelevant attribute analysis. This study confirms that ensemble algorithms can be implemented for the development of more effective and accurate phishing detection systems. XGBoost is recommended as the primary algorithm in detecting phishing threats in cybersecurity applications, thanks to its ability to handle large and complex data.