Jan Nunez is a security researcher for a Fortune 50 company where he is primarily responsible for application security across web technologies, mobile applications and mainframe systems. He has presented his research at several industry conferences, including ThotCon, ShmooCon and BSides Las Vegas. His latest project is the UFO Reporting System - a test bed for AI security and other emerging technologies, and not, disappointingly, a hotline for reporting UFOs. When he's not destroying his posture over a keyboard, he enjoys dancing, archery and practicing martial arts.
Artificial Intelligence is raging through cyberspace, but can we handle the ride? It's time for a systems check. Join security researcher Jan Nunez as he uncovers the attacks AI/ML systems are vulnerable to and the controls currently available to mitigate them.
No prior experience with AI/ML is required. The talk will start with an overview of machine learning, highlighting its benefits and the events that led to its rise in popularity. We'll then shift our focus to practical applications with live demos using the UFO Reporting System, an application designed to showcase cutting-edge attacks on AI systems and other emerging technologies.
We will cover several attack scenarios, including remote code execution through infected models, creating adversarial images using gradient descent, and fancy prompt injections in a LLM used for tracking UFO encounters.