KEYNOTE // Adapting to the brave new world of computer security
Bruce Dang, Calif.io
KEYNOTE // Adapting to the brave new world of computer security
Abstract
AI capabilities in difficult tasks such as bug discovery, exploitation, and reverse engineering have been improving at a breakneck speed. The cost of a Linux kernel exploit is less than $20. OpenBSD kernel exploits are no longer rare unicorns. Does this mean computer security is a "solved" problem for AI? What are the remaining problems? In this talk, I will share our experience applying frontier models to various problems in computer security and discuss the successes and failures. In addition, I will discuss future challenges and what we need to do to prepare for them.
Bruce Dang
Bruce Dang is a failed recluse who has been dabbling with computer for a few years. He started his security career in the primordial days of the Microsoft Security Response Center (MSRC) when it averaged three or 4 Windows bugs per month. Along the way, he analyzed Stuxnet and wrote a popular children's book called Practical Reverse Engineering with some friends. After realizing the impermanence of Windows, he seeked out the elders at Apple and was re-educated in computer security. As a novice mendicant at Apple, he learned the real challenges of privacy, security, and debugging without attachment or anger. His single-minded pursuit was interrupted by the iconoclasts at Calif.io; they convinced him that AI is the only true path to enlightenment. Since then, he has been studying the endless sacred scriptures of Claude. He has not reached AGI.
