AI-Powered Phishing Website Detection using Machine Learning and Browser Extension Integration

Cybercrime has many forms, and phishing is one of the most common types. There were almost 989,000 phishing incidents in the first quarter of 2024 alone! Because conventional defenses rely on blacklists to catch phishing URLs, they cannot detect phishing attacks that use new or different types of phishing URLs. This leaves millions of users at risk of having their data stolen, their identities compromised, or their money stolen as a result of phishing. This study presents a new phishing detection system called PhishGuard. PhishGuard is a browser extension that runs on Google Chrome and uses machine learning to classify each website as either a phishing website or a legitimate website in real-time. PhishGuard collects various URL-based and domain-based features of websites (e.g., URL length, special character count, HTTPS status, redirection type, etc.) and input these features into an optimized ensemble protocol that is based upon a combination of multiple classification algorithms. PhishGuard was tested against a number of established methods of phishing detection, including Logistic Regression, Decision Tree, and Support Vector Machines. The results demonstrate that PhishGuard outperformed these approaches in terms of detection accuracy. PhishGuard achieved an accuracy rating of 96.75% compared to Logistic Regression at 88.45%, Decision Tree at 91.20%, and Support Vector Machines at 90.35%. PhishGuard provides users with instant alerts after visiting a phishing website, maintains a browsing history log, and works automatically without requiring any manual user involvement. PhishGuard is an innovative, lightweight, scalable, and practical anti-phishing tool that can be deployed in any browser environment, with clear capabilities for future integration of enhanced deep learning processes and real-time threat intelligence.