Surakshit Web: An AI-Powered Phishing URL Detection System

Abstract

Phishing attacks have become a significant cybersecurity threat, exploiting users' trust to steal sensitive information through deceptive URLs. Traditional detection methods often fail to keep up with evolving phishing techniques. This research proposes Surakshit Web, an AI-powered phishing URL detection system leveraging deep learning and optimization techniques. The system integrates Convolutional Neural Networks (CNNs) and a Genetic Algorithm (GA)-optimized Multi-Layer Perceptron (MLP) to enhance detection accuracy. A comprehensive dataset of malicious and benign URLs was used, incorporating advanced feature extraction and data pre-processing techniques to improve model performance. The results indicate that the GA-optimized MLP outperforms CNN, achieving a 96.12% accuracy compared to CNN’s 93.78%, demonstrating its effectiveness in identifying phishing URLs. This research highlights the potential of evolutionary algorithms in optimizing deep learning models for cybersecurity applications.