This project investigates market-driven spectrum access and management approaches that leverage artificial intelligence (AI) to enhance radio spectrum allocation, sensing, and market optimization. The work facilitates evolution from current radio spectrum management strategies to more dynamic and efficient methods, thereby increasing the overall utility and efficiency of the radio spectrum which is a key resource for all sectors of modern society. The new approaches investigated in this project rely on private sector band managers, who dynamically allocate spectrum resources while ensuring compliance and mitigating interference. Band managers must contend with strategic behavior by market participants, who for example may share incomplete or incorrect information, and must handle attacks by adversarial users. This research addresses these challenges by developing AI-driven mechanisms that balance efficiency, security, and stability. Deployment of the new market-driven approaches could significantly enhance the spectrum available to and hence the capacity of next-generation wireless communication systems and other spectrum dependent systems. This research comprises three integrated thrusts considering different aspects of a future robust AI-powered market-driven spectrum system. Thrust 1 develops new learning-based spectrum allocation mechanisms that integrate multi-armed bandits with auction strategies to optimize spectrum sharing among strategic users. Thrust 2 focuses on