AI-Driven Detection of Adversarial Attacks in Post- Quantum Cryptographic Systems

The rise of quantum computing threatens traditional cryptographic systems, necessitating the development of post - quantum cryptographic (PQC) algorithms. However, these algorithms remain susceptible to adversarial attacks, including chosen ciphertext attacks (CCA), side - channel attacks, and machine learning - induced adversarial threats. To address this, we propose an AI - based adversarial attack detection framework that enhances PQC security by employing deep learning and anomaly detection techniques. Our approach utilizes Graph Neural Networks (GNNs) and transformer - based models to identify cryptographic perturbations in real - time. The framework continuously monitors security metrics, analyzing attack vectors such as timing variations, side - channel leakages, and adversarially modified ciphertexts. This study contributes to advancing quantum - resilient cryptographic security and will be presented at international cybersecurity and AI conferences.