Explore how artificial intelligence agents can autonomously identify complex logic-based vulnerabilities in APIs through this 37-minute conference talk. Learn about training reinforcement learning agents to navigate applications and discover security flaws that traditional scanning methods might miss. Discover the implementation of the Marvin framework with vulnerable APIs as a testing ground for these AI-driven security assessment techniques. Examine both the promising capabilities and current limitations of using machine learning approaches for automated vulnerability discovery. Gain insights into the intersection of artificial intelligence and cybersecurity, specifically focusing on how autonomous agents can enhance API security testing methodologies.