Explore the critical vulnerabilities and deceptive tactics that can compromise AI systems in this 11-minute conference talk from All Things Open's machine learning and AI track. Delve into the methods attackers use to exploit artificial intelligence models and discover how these systems can be fooled through various techniques. Learn about the importance of building robust defenses against AI manipulation and understand the evolving landscape of AI security threats. Gain insights into best practices for protecting AI systems from adversarial attacks and developing more resilient machine learning models that can withstand sophisticated deception attempts.