Lessons Learned from Building and Defending LLM Applications

Explore real-world insights and practical lessons from building and defending Large Language Model (LLM) applications in this 27-minute conference talk from DEF CON 32. Gain firsthand knowledge about critical vulnerabilities encountered during development, innovative defense mechanisms implemented, and essential strategies for strengthening AI application security. Learn actionable approaches to navigate the complex landscape of AI security, drawing from direct experience in developing and protecting LLM systems in production environments. Master practical techniques for enhancing the robustness of AI applications while understanding common pitfalls and their solutions in LLM implementation and security.