Securing Generative AI: Threats Old and New
Explore the security implications of generative AI in this 26-minute conference talk from BSidesCharm 2024, delivered by AI Security Researcher Adam Swanda. Gain essential insights into AI security concepts, with particular emphasis on Large Language Models (LLMs), and discover how traditional cybersecurity approaches can be adapted for emerging AI systems. Learn from Swanda's decade-plus experience in cybersecurity and threat intelligence as he breaks down the unique security challenges and solutions in AI adoption. Benefit from the expertise of the creator of "Vigil," an open-source Python library designed for LLM defense against prompt injection and jailbreak attempts, while developing a comprehensive understanding of the evolving AI security landscape.