Real-Time AI/ML-Based Phishing Detection and Prevention System

Including irrelevant or highly correlated features increases model complexity without improving performance. This may lead to overfitting and unstable predictions.     II.           6. Lack of Dataset Diversity Training models on limited or outdated datasets results in poor generalization. Many systems fail when exposed to new phishing campaigns, languages, or regional attack patterns.     III.         7. Absence of Explainability Black-box models without explainability mechanisms make it difficult to understand decisions, reducing trust and complicating regulatory compliance and incident analysis.     IV.           8. Ignoring User Behavior Signals Many implementations focus only on URL or content analysis while ignoring behavioral indicators such as click timing, mouse movement, or session anomalies, which can significantly enhance detection accuracy.     V.                G. Improper Evaluation Metrics Relying solely on accuracy is misleading, especially for imbalanced datasets common in phishing detection. Metrics such as precision, recall, F1-score, and ROC-AUC are often overlooked.     VI.          10. No Feedback or Learning Loop Systems that do not incorporate user or analyst feedback fail to improve over time, missing an opportunity to adapt to new phishing tactics.  

[1]          S. Garera, N. Provos, M. Chew and A. D. Rubin, A framework for detection and measurement of phishing attacks, in Proceedings of the 2007 ACM Workshop on Recurring Malcode, 2007, pp. 1–8. ResearchGate [2]           A. U. Rehman, I. Imtiaz, S. Javaid and M. Muslih, Real-Time Phishing URL Detection Using Machine Learning, Security, vol. 107, 2025. MDPI [3]          D. T. Mosa, Machine Learning Techniques for Detecting Phishing URL, Journal of Information Security and Applications, 2023. ScienceDirect [4]       S. H. Ahammad Shaik et al., Phishing URL detection using machine learning methods, Advances in Engineering Software, 2022. ResearchGate   [5]          N. Nagy, Phishing URLs Detection Using Sequential and Parallel Machine Learning Approaches, Sensors, vol. 23, no. 7, 2023. MDPI [6]          S. K. Kapan, Improved Phishing Attack Detection with Machine Learning, Applied Sciences, 2023. MDPI [7]          A. Almujahid et al., Comparative evaluation of machine learning and deep learning algorithms for phishing detection, PMC Scientific Reports, 2024. PMC [8]          M. A. Tamal et al., Dataset of suspicious phishing URL detection, Frontiers in Computer Science, 2024. Frontiers [9]        J. L. Wilk-Jakubowski, Machine Learning and Neural Networks for Phishing Detection: A Comprehensive Review, Electronics, 2025. MDPI [10]        A. UZ. Asif, Machine Learning-Based Phishing Detection Using URL Features: A Systematic Survey, Proc. URL-Based ML Security, 2023. ACM Digital Library [11]        R. Dubey, A. M. Tripathi, A. Srivastava and S. Singh, Phishing Detection System: An Ensemble Approach Using Character-Level CNN and Feature Engineering, arXiv preprint, Dec 2025. arXiv [12]        S. Baskota, Phishing URL Detection using Bi-LSTM, arXiv preprint, Apr 2025. arXiv [13]        S. Aslam et al., AntiPhishStack: LSTM-based Stacked Generalization for Optimized Phishing URL Detection, arXiv preprint, 2024. arXiv [14]        A. Le, A. Markopoulou and M. Faloutsos, PhishDef: URL Names Say It All, arXiv preprint, 2010 — foundational work on lexical features for phishing detection. arXiv [15]   “Malicious URL Detection using Machine Learning: A Survey”, by D. Sahoo, C. Liu and S. C. H. Hoi, ACM Computing Surveys, 2017 — survey background on ML approaches.