Tinker Tailor LLM Spy - Investigate and Respond to Attacks on GenAI Chatbots

Learn to investigate and respond to security incidents targeting generative AI chatbots in this 39-minute Black Hat conference talk. Discover how to prepare for and handle the inevitable security breaches when your company's LLM-powered virtual agents start behaving inappropriately, leaking customer PII, or falling victim to prompt injection attacks. Explore new methodologies for isolating attacks, gathering critical incident information, and identifying root causes using specialized AI defense tooling and LLM guardrails. Master the connection between traditional cybersecurity investigation procedures and the unique threat landscape of GenAI systems. Gain practical knowledge about the specific vulnerabilities that plague LLM-based chatbots and understand why standard security measures often fall short. Develop a comprehensive incident response playbook tailored specifically for GenAI chatbot security breaches, including preparation steps to implement before your chatbot responses become a viral security nightmare. Walk away with actionable strategies for building resilient AI systems and the investigative skills needed to handle this emerging class of security incidents effectively.