Juan Andrés Guerrero-Saade (JAGS), VP of Intelligence & Security Research at SentinelOne, delivers a keynote arguing that cybersecurity is at the end of its experimental era. Decades of complexity layered onto non-standardized stacks have made systems hard to manage with human attention alone. He proposes that LLMs offer a new source of scalable, cheap evaluative power — a 'lossy compression of human knowledge' — that can help defenders assess, prioritize, and act without relying solely on scarce human expertise. Drawing on cybernetics, he advocates moving beyond purely adversarial design toward systems where human expertise and AI work together, pointing toward a more standardized, automated, and sustainable security industry.
141 Impressions