PhishGuard: A Cloud-Based Machine Learning Approach to High-Accuracy Phishing URL Detection

Phishing attacks represent one of the most significant cybersecurity threats in the digital age, causing substantial financial losses and compromising sensitive information. This paper presents PhishGuard, a comprehensive cloud-based phishing detection system that combines machine learning with external threat intelligence to detect malicious URLs with high accuracy. The system employs a multi-layer verification approach using ensemble methods including Random Forest and XGBoost algorithms, trained on extensive datasets from PhishTank and legitimate URL sources. Our approach achieves 95.8% detection accuracy while maintaining processing times under 15.2ms per URL. PhishGuard implements a scalable cloud-native architecture supporting real-time URL analysis through RESTful APIs, enabling seamless integration with browser extensions, email clients, and enterprise security systems. The system addresses critical gaps in existing solutions by providing non-intrusive, scalable, and easily deployable cybersecurity protection. Experimental validation across 100,000 URLs demonstrates robust performance with minimal false positives, making it suitable for practical deployment in production environments. Index Terms—phishing detection, machine learning, cloud computing, cybersecurity, URL analysis, Random Forest, XGBoost, ensemble methods, threat intelligence