As adversarial AI systems are constantly evolving, and new attacks are being developed, they create a serious threat for cybersecurity, especially for the detection of complex, adaptive, and unseen threats. This paper introduces a quantum augmented AI system to extract knowledge from vast amounts of cybersecurity data and to build knowledge graph dynamically for enhancing adversarial threat detection. The framework takes advantage of recent developments in quantum utility by mapping high-dimensional and complex security data into quantum Hilbert spaces using Quantum Support Vector Machines (QSVM) and Quantum Neural Networks (QNN) to uncover hidden relationships and polymorphic attack patterns. The proposed approach leverages the ability of quantum machine learning algorithms to detect subtle adversarial behaviors, anomalous interactions, and zero-day vulnerabilities that could escape traditional machine learning defenses, by integrating data mining and knowledge graph analytics with AI. The framework offers a scalable approach to harnessing cybersecurity intelligence from large and diverse data sources and improves the security systems' strength and flexibility. The experimental analysis shows that the use of quantum-assisted models can enhance the representation of threats, pattern recognition, and proactive cyber defense, making quantum machine learning a potential direction for next-generation adversarial cyber defense solutions.