Year: 2025
Type: conference-paper
Abstract: Phishing remains a major cybersecurity concern, where attackers steal sensitive data. Traditional methods of detection, such as blacklists and rule-based systems, tend to fail to detect new or emerging threats. To address this gap, we propose a machine learning–based browser extension that detects phishing websites in real time. The extension silently operates in the background of a user's web browser, examining factors including URL structure, webpage content, domain legitimacy, and visual signals to accurately classify them. Fundamentally, at its center is an XGBoostclassifier that has been trained on a well-filtered and varied dataset. When a phishing attempt is recognized, users are promptly notified with no perceptible interruption to their browsing session. The system achieved 95% accurate during testing and has good precision and recall rates that indicate its robustness. This lightweight and efficient tool not only secures users against online attacks but also learns to keep up with evolving attack trends. In the future, we intend to investigate deep learning methods and add support to additional browsers, solidifying user security in cyberspace.
Source: AIJR Proceedings
Authors: Shyni Shajahan, Teena George, P V Jithi, K M Snehamol, Sanjana Krishna, Sanjaly Krishna, Nagul Jagadish
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4415247213