This course continues the successful workshop on AI adversarial attacks organised in collaboration with Transmediale festival. Rather than studying AI step-by-step, the course focuses on AI vulnerabilities and limitations in order to understand and disclose its inner workings. As in its traditional definition, the practice of "hacking" is understood as a tool of knowledge to discover the structure of a machine, while the adjective "adversarial" is taken as a metaphor to describe new general strategies of data activism and privacy protection. The issue of AI vulnerabilities will be discussed in its broader cultural, economic and political implications. Full description forthcoming.