Explore critical security and privacy challenges in AI systems through this 28-minute conference talk from the Cloud Security Alliance AI Summit. Dive into adversarial machine learning concepts and understand how AI models can be compromised in critical applications like healthcare, finance, autonomous vehicles, and cybersecurity. Learn about a comprehensive taxonomy of AI attacks and mitigations developed with NIST, examine real-world case studies of attacks on generative AI and AI-enabled systems, and discover defensive strategies for protecting AI models against adversarial threats. Master practical approaches for building more resilient AI systems while gaining valuable insights into current open challenges in AI security design and implementation.