ABOUT THIS EPISODE
In this episode of Unprompted, Chunga sits down with Diana Kelley, Noma Security CISO and longtime AI security leader, to discuss the simple methodology she uses to demystify the constantly changing landscape of AI. From large language models, to transformers and threat modeling, Diana helps Chunga breakdown why AI may feel magical, but isn't.
Together, they explore what AI actually does (and doesn't do), why it doesn't think, how security leaders should approach AI risk, and where hype gives way to practical reality. Along the way, they tackle over-reliance on AI, agentic systems, and what CISO's need to know as AI becomes further embedded into modern infrastructure.
Chapters
00:00 - The origin of "AI isn't magic, its math"
05:00 - Machine learning, transformers, and why ChatGPT feels different
12:00 - Why AI doesn't actually think
18:00 - Will AI ever be sentient?
22:00 - Over-reliance on AI and security risk
27:00 - aI as an enhancer of human work
32:00 - Agentic AI, threat models, and the myth to debunk
To learn more about how you can secure your AI everywhere, go to noma.security
