Tinker Tailor LLM Spy - Investigate and Respond to Attacks on GenAI Chatbots - 04

Learn how to investigate and respond to security incidents targeting generative AI chatbots in this 56-minute conference talk from BruCON Security Conference. Discover the unique threats facing GenAI chatbots that leverage large language models for customer engagement, including scenarios where virtual agents respond inappropriately or leak customer personally identifiable information. Explore traditional investigation and response procedures adapted for the GenAI environment, moving beyond basic awareness of prompt injection attacks and LLM guardrails to practical incident response skills. Master new methods for isolating attacks, gathering critical information, and identifying root causes using AI defense tooling and LLM guardrails. Gain a comprehensive playbook for investigating and responding to GenAI incidents, along with essential preparation steps to implement before your company's chatbot responses create negative publicity. Understand how to bridge conventional cybersecurity practices with the emerging challenges of securing generative AI systems in production environments.