SN 1028: AI Vulnerability Hunting - The End of Jailbreaking
Security Now (Video)
00:00
Exploring Vulnerabilities with LLM Technology
This chapter introduces the minimalistic LLM tool for interacting with large language models while highlighting a significant zero-day vulnerability in the Linux kernel's SMB implementation. It explores the intricacies of vulnerability hunting, especially focusing on how reference counting and concurrent programming affect memory management. The discussion demonstrates how LLMs, like OpenAI's O3 model, can uncover critical security flaws through effective automation in vulnerability research.
Transcript
Play full episode
