Episode Details

Delve into the critical security vulnerabilities of Artificial Intelligence, exploring the dangerous world of prompt injection, leaking, and jailbreaking as highlighted in SANS' Critical AI Security Controls and real-world adversarial misuse of generative AI like Gemini by government-backed actors. Understand how malicious actors attempt to bypass safety controls, extract sensitive information and manipulate LLMs for nefarious purposes, drawing insights from documented cases involving Iranian, PRC, North Korean, and Russian threat actors. Learn about the offensive techniques used and the ongoing challenge of securing AI systems,