In the evolving landscape of cybersecurity, the increasing complexity and frequency of attacks demand intelligent systems capable of proactive threat detection and adaptive reasoning. This research presents an Adaptive AI-Driven Knowledge Graph Framework designed to enhance proactive threat hunting and dynamic cyber risk assessment. The proposed framework integrates knowledge graphs with adaptive artificial intelligence to represent, learn, and reason over heterogeneous threat data. By dynamically correlating indicators of compromise, behavioral attributes, and contextual relations, the system uncovers latent attack patterns that traditional methods often overlook. The adaptive AI layer continuously refines its knowledge through feedback-driven learning, enabling real-time response and improved situational awareness. Experimental evaluations demonstrate that this framework significantly improves detection accuracy, correlation efficiency, and risk prediction reliability compared to conventional models. The study highlights the potential of combining semantic graph intelligence with adaptive analytics to create resilient, explainable, and self-evolving cybersecurity ecosystems capable of addressing emerging threats in complex network environments.

Keywords: AI-driven cybersecurity, knowledge graphs, threat hunting, cyber risk assessment, adaptive intelligence