About Session
AI systems are being deployed at unprecedented speed — but most organizations have never tested them for security vulnerabilities. 73% of enterprises running AI models have no red team process in place. This session pulls back the curtain on how attackers exploit large language models and autonomous agents, and what defenders can do about it. Attendees will learn about red teaming — the practice of simulating real-world attacks to identify weaknesses before malicious actors can exploit them — along with techniques like prompt injection, jailbreaking, multi-step attack chains, and behavioral fingerprinting. Featuring a live walkthrough of an AI red teaming platform that tests 280 attack techniques across 25 categories, this presentation gives security leaders, engineers, and executives the practical knowledge to start building their AI security program today. Whether you’re deploying chatbots, copilots, or autonomous agents — if you’re running AI, you need to be red teaming it.
