The rapid evolution of cyberattacks, coupled with the increasing capacity of computing environments and the emergence of artificial intelligence (AI), has significantly complicated the security landscape. While existing studies largely emphasize improving AI model performance for individual cybersecurity tasks, this survey shifts the focus toward operationalizing the deployment rationale and understanding when, where, and why different AI paradigms should be deployed, the capabilities they offer; and the challenges that must be addressed to enable trustworthy and effective real-world cyber defense. This paper aims to provide researchers and practitioners with a comprehensive reference for understanding the evolving role of AI in cybersecurity and the challenges that must be addressed to develop trustworthy and resilient AI-driven cyber defense systems. In this paper, we propose a structured taxonomy to organize various dimensions of AI-driven cybersecurity; review them critically; and finally, discuss key challenges, open problems, and emerging trends.