The integration of Internet of Things (IoT) devices with green computing cloud infrastructures has enabled scalable and energy-efficient services but has also amplified the threat of sophisticated phishing attacks. Traditional detection techniques, including signature-based and conventional machine learning models, often fail to adapt to the evolving tactics of phishing campaigns in dynamic cloud environments. This paper proposes a novel phishing detection framework based on the Deep Deterministic Policy Gradient (DDPG) algorithm, specifically designed for IoT-enabled green cloud infrastructures. The framework introduces three key innovations: a dynamic feature extraction module that adapts to changing phishing patterns in cloud traffic, a real-time anomaly detection system optimized for energy efficiency, and an adaptive policy optimization mechanism that enhances detection accuracy (96.8%), precision (95.4%), recall (94.8%), and F1-score (95.1%) over time. Experimental results demonstrate that the DDPG-based approach significantly reduces false positive rates by up to 16% compared to traditional methods while maintaining low computational overhead, making it wellsuited for sustainable and secure cloud computing environments.