Systematic Review of Issues in Identification and Detection of Malicious URLs Using GNN

The threat of cybersecurity continues to rise as malicious URLs have become one of the most critical threats for phishing attacks, malware distribution and identity theft. But the phishing URLs exploit trust built in users to get their sensitive data and cause significant financial losses, operational disruptions and damage to an individual's and organizations' reputation. Current detection approaches, including Machine Learning (ML) techniques are based on traditional means and insufficient against the modern and evolving phishing techniques. URL features such as structure, domain reputation, and content are analyzed using ML models such as Support Vector Machine (SVM), Random Forest, and Neural networks (NN) to increase detection accuracy. These models suffer from scalability, real time adaptability, and zero-hour phishing attacks detection which takes advantage of newly discovered vulnerabilities in software or systems before a patch or fix is available. This paper proposes an innovative approach for real time, user friendly malicious URL detection by integrating the Graph Neural Networks (GNNs) into a chatbot interface. Advanced approaches like Graph Neural Networks (GNNs) and Transformers can improve detection performance by integrating contextual relationships and semantic analysis. In addition to this, it also studies URL masking, false positives and the complexity of feature engineering. This paper compares traditional and emerging methods to show how these upcoming technologies can improve the capabilities of detection to strengthen the security and trust of users on the digital platforms.