Cybersecurity underpins the lives of ordinary people-their safety, work, health, and entertainment. Yet despite its importance, cybersecurity is often approached in a reactive manner-taking corrective actions to "patch" vulnerabilities after they are detected or exploited. In the absence of fundamental improvements in formal processes and methods, the same or similar problems can recur. In contrast, a scientific vision for security and privacy should be proactive. The focus of this TOIT special section is to provide a forum to discuss and advance security and privacy by means of Artificial Intelligence (AI) approaches. Approaches that are intelligent and self-adaptive are crucial to deal with the complexities of effectively protecting sensitive assets in all security-critical domains. This is where research from the AI community can make a difference in security and privacy. Specifically, AI can help address a long-standing problem, namely, that security and privacy are attempted to be inserted as an afterthought, which is rarely adequate. We identify the following key areas in AI, and review advancements in such areas that would help solve fundamental hard problems regarding security (CyBOK 2017) and privacy (Such 2017). Normative models: Secure and privacy-aware governance of sociotechnical systems can only be achieved by bridging the divide between technical solutions to security and privacy such as access control systems and the human and social factors associated with the users of such systems. Development of unified computational models for the social and the technical tiers of sociotechnical systems as well as their verification are therefore of utmost importance.