Red Team | The Current State and Future of AI's Role in Vulnerability Research and Offensive 🎙️ Stephen Sims, Research Fellow, SANS Institute 📍 Presented at SANS Hack & Defend Summit 2025 The speed at which AI security products are being released is unlike anything we've seen in the past. The opportunities to weaponize this technology for offensive operations are endless! In my research I've been working on using agents to perform duties such as binary and patch diffing, binary vulnerability research and fuzzing, and novel techniques to train and empower agents to identify 0-day vulnerabilities in a large number of web applications. The success has been greater than I had imagined. Join me in this talk where we will take a look at each of these areas and the methodology behind model training, hallucination reduction, memory management, and other relative development. You will leave this talk with a better understanding as to how AI and automation is being used today and where things are headed.