New findings from Pentera’s AI Security & Exposure Survey 2026 reveal a critical blind spot: while AI adoption is widespread, most organizations lack the skills to secure it, have limited visibility into where AI is deployed, rely on legacy security controls not designed for AI-driven workflows and - most concerning - still have unclear or undefined approaches to AI security testing. As a result, AI-driven connections across data, identity, and infrastructure remain exposed in ways attackers can exploit. Join Ryan Dorey, Director of Technical Advisors at Pentera, and Chris Cochran, Field CISO & VP of AI Security at SANS, for a deep dive into the benchmark data. During the Session, We Will: Break down the benchmark methodology and key findings from 300 CISOs Share examples of how AI security gaps translate into exploitable risk Show how adversarial testing helps security teams validate AI risk