Published Paper Details:

Phishing attacks remain one of the most persistent and evolving threats to internet security, causing significant financial losses and compromising sensitive data worldwide. To address this challenge, Web Integrity Shield presents an intelligent, real-time phishing detection system aimed at safeguarding users by proactively identifying malicious websites before harm occurs. By combining traditional URL-based heuristics with advanced machine learning, the system provides a proactive layer of defense that evolves alongside emerging phishing tactics. The proposed solution employs hybrid AI architecture integrating multiple technologies for accuracy and scalability. Initially, a machine learning model is trained in Python using a comprehensive dataset containing URL-based heuristic features. To further enhance detection, the system leverages Selenium WebDriver for automated, headless browser analysis within a secure, sandboxed backend. This enables dynamic extraction of features from the website's DOM structure, JavaScript execution patterns, and visual layout. The enriched feature set is then processed by the trained model, serialized into ONNX format, and deployed through a Java Spring Boot backend. A lightweight JavaScript browser extension functions as the client, performing quick feature extraction and querying the backend API for instant predictions. This modular architecture ensures scalability, high accuracy, and separation of data science and deployment. The system delivers real-time protection against evolving phishing threats while remaining efficient and user-friendly