The relationship between AI and cybersecurity is of conflicting nature. While threat detection and containment can be perfected by the use of new AI tools and methodologies, the current data deluge and the increasing complexity of information and communication technologies ( ICT ) make almost impossible to properly protect them without the guidance of automatic decision making solutions. However, an excess of confidence in such solutions can compromise security and enclose safety risks for information systems. Moreover, neglecting the treatment of these risks could undermine the different modalities of governance that configure some pillars of our democracy. Fundamental citizens ’ rights such as privacy or accountability in public procurement are major components of the conundrum related to the alignment of technological possibilities with ethical, legal and normative regulations. This chapter summarizes the CSIC approach to tackle the above challenges in the crossed domain of AI and cybersecurity, underlining the need for an integral strategy for the deployment of secure and safe AI systems. This approach encompasses the whole stack associated with the design and implementation of AI solutions, ranging from the hardware to the application layer, and considering the theoretical underpinnings to adequately bridge AI functionality and cybersecurity requirements.
