PHISHING DETECTION IN WEBSITES

Abstract

This project proposes a phishing detection system that functions as a real-time Chrome browser extension powered by Artificial Intelligence. Traditional methods relying solely on blacklists or URL structure often fail to catch cleverly disguised phishing websites. To overcome this, our approach introduces a hybrid detection model that first checks websites against a known phishing database for instant blocking. If no match is found, the system extracts both URL features and visible text content from the page. Using ensemble learning for URL analysis, the system evaluates threats with deeper context. The ensemble model generates the final result by combining evaluations from both machine learning and deep learning models. This two-layer detection significantly improves accuracy and reduces false positives by going beyond rule-based logic. The result is a faster, smarter, and more reliable tool for protecting users during everyday browsing.
Phishing is one of the most common and
dangerous cyberattacks, where at tackers
create fake websites that mimic legitimate
ones to steal sensitive information such as
usernames, passwords, and banking
details. With the rapid growth of online
services, phishing attacks have become
more sophisticated, making it difficult for
use rs to distinguish between genuine and
malicious websites. Traditional security
methods such as blacklists and rule based
systems are no longer sufficient to detect
newly generated phishing websites.