Phishing Detection Using a Convolutional Neural Network Model on Website URLs
Abhishek Aman Swamy, Dinesh Kumar · 2024
According to recent statistics, most business activity is now conducted online via the internet and through websites. This shift to online commerce has led to risk of exposure to cyber threats, such as phishing. Phishing detection remains a challenge, with traditional blacklisting methods struggling to keep pace with the exponential growth of malicious websites. As a result, phishing attacks have become more frequent and sophisticated and therefore requires newer and smarter ways of detection. This paper proposes a two-step, computer vision-based technique for phishing detection. In the first step, URLs are converted into images. Then, a Convolutional Neural Network (CNN) is employed to analyse these images and classify them as phishing or legitimate. Our experiments demonstrate promising results, indicating that the proposed method outperforms traditional approaches in phishing URL detection, such as blacklisting. This finding highlights the potential of utilizing URL-to-image conversion and CNNs for effective phishing website classification.