This research introduces “Cyber Sentinel,” an AI-based system enhancing phishing website detection through dataset fusion from sources like Phishtank, Mendley, Kaggle, and the UCI datasheet. Models, including random forest, XGBoost, SVM, and logistic regression, are trained on a unified dataset using ensemble learning techniques for real-time website classification. Cyber Sentinel demonstrates high accuracy, with XGBoost and logistic regression models achieving rates of 96.3 and 96.5%, respectively. Evaluation metrics such as precision, recall, and F1-score validate its proficiency in identifying phishing websites. Real-world testing confirms its adaptability to new phishing patterns. The research outlines future directions, emphasizing continuous model training, deep learning integration, and user collaboration for ongoing refinement. Cyber Sentinel stands as a robust AI-driven solution for fortifying online security against evolving phishing attacks, contributing significantly to advancements in cybersecurity measures.