Zero-Day Attack Detection Using Generative Artificial Intelligence and Graph Neural Networks

Zero-day attacks exploit previously unknown vulnerabilities, posing a severe threat to modern digital infrastructures. Traditional signature-based intrusion detection systems are unable to identify these attacks, making timely detection a major challenge (Bilge & Dumitras, ۲۰۱۲). This study proposes a novel hybrid framework that integrates Generative Artificial Intelligence (GenAI) with Graph Neural Networks (GNNs) to detect zero-day attacks in network traffic. The generative component models normal and malicious traffic patterns while synthesizing realistic attack scenarios, enabling the system to recognize unseen threats (Goodfellow et al., ۲۰۱۴). Concurrently, the GNN captures structural dependencies within network communication graphs, enhancing anomaly detection in complex network topologies (Wu et al., ۲۰۲۱). Experimental evaluations on benchmark datasets, including CICIDS۲۰۱۷ and NSL-KDD, demonstrate that the proposed model achieves higher detection accuracy and lower false-positive rates compared to conventional machine learning and deep learning baselines. These results indicate that integrating GenAI with GNNs provides a scalable and effective solution for zero-day attack detection, improving the resilience of network security systems. The framework offers a practical approach for real-time detection and holds potential for adoption in critical digital infrastructures to mitigate evolving cybersecurity threats.

کلیدواژه ها:

Zero-Day Attacks ، Intrusion Detection ، Generative AI ، Graph Neural Networks ، Cybersecurity