Phishing attacks remain a significant cybersecurity threat in the digital landscape, leading to the development of defense mechanisms. This paper presents a thorough examination of Artificial Intelligence (AI)-based ensemble methods for detecting phishing attacks, including websites, emails, and SMS. Through the screening of research articles published between 2019 and 2023, 37 relevant studies were identified and analyzed. Key findings highlight the prevalence of ensemble methods such as AdaBoost, Bagging, and Gradient Boosting in phishing attack detection models. Adaboost emerged as the most used method for website phishing detection, while Stacking and Adaboost were prominent choices for email phishing detection. The majority-voting ensemble method was frequently employed in SMS phishing detection models. The performance evaluation of these ensemble methods involves metrics, such as accuracy, ROC-AUC, and F-score, underscoring their effectiveness in mitigating phishing threats. This study also underscores the availability of credible open-access datasets for the progressive development and benchmarking of phishing attack detection models. The findings of this study suggest the development of new and optimized ensemble methods for phishing attack detection.