Abstract
This manuscript surveys the transformative convergence of Generative AI (GenAI) and Explainable AI (XAI) in cybersecurity, addressing computational democratization and evolving threats like polymorphic malware 2.0, LLM-chaining, deepfakes, and adversarial poisoning. Through systematic review of 2022-2025 literature, it analyzes GenAI applications in SOCs such as dynamic playbooks, LLM threat hunting, GAN synthetic data—and critiques XAI limitations (e.g., SHAP latency), proposing neuro-symbolic hybrids and federated frameworks. Hardware innovations like AI-driven Trojan detection and PUFs are evaluated alongside taxonomies for unified paradigms. Findings reveal enhanced detection accuracy and interpretability, yet gaps in real-time scalability persist. Future directions emphasize causal reasoning, adversarial robustness, and regulatory compliance for trustworthy cyber defense.

This publication has 33 references indexed in Scilit: