A Systematic Literature Review of Intelligent Anomaly Detection and Threat Analysis Using LLMs in Quantum-Aware Cyber Systems

Abstract

AI-driven cyber attacks are growing in number, precision, and complexity. Large Language Models (LLMs), which are AI systems capable of understanding and generating human language, and quantum computing are simultaneously reshaping the cybersecurity threat landscape, creating an urgent need for intelligent, quantum-aware cyber defense. This paper presents a Systematic Literature Review (SLR) following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework, yielding 74 peer-reviewed studies from 2015 to 2026 across five themes: AI-based threat analysis, LLM applications in cybersecurity, anomaly detection and Advanced Persistent Threat (APT) identification, deep-learning intrusion detection systems, and post-quantum cryptography (PQC). The review finds that LLMs are deployed not only to classify attacks but to reason about them reading logs, generating detection rules, and supporting incident response. Six critical gaps are identified: scarce public APT datasets (G1), hallucination risk in security-critical LLM deployments (G2), absence of lightweight PQC for IoT (G3), no unified quantum-aware AI framework (G4), insufficient real-time edge-detection latency guarantees (G5), and unverified LLM robustness against adversarial prompt injection in production pipelines (G6).