Cybersecurity's AI vs. AI Trap | J.R. Rao (CTO, Security Research, IBM)

Listen to full episode:

Summary: IBM Fellow and CTO of Security Research J.R. Rao joins Anant and Ed to talk about what AI means for security. J.R. walks through how every major platform shift, from mainframes to AI, redraws the boundaries of trust, then digs into what happens when AI systems start operating with real autonomy and where that creates new exposure. The conversation also covers how AI can strengthen security work while introducing harder problems of its own, and closes with a look at what's coming next, including the intersection of AI and quantum computing.

Chapters:

00:00 - Hosts say hello and establish the episode topic

01:37 - Discussion on model announcements and autonomous loops

05:11 - Emerging headlines and industry caution on AI capabilities

07:49 - Introduction of J.R. Rao and his security insights

09:18 - J.R. on security innovation driven by industry platforms

12:25 - Emergent behaviors of AI agents and security implications

15:23 - Security architecture and defense in depth for AI

17:14 - The importance of rigorous security controls and culture

20:56 - Evolution of security architecture from perimeter to zero trust

24:38 - Challenges of AI autonomy and non-determinacy

30:09 - Using AI to improve security and fix vulnerabilities

33:21 - Open source vulnerabilities and IBM's Lightwell project

36:48 - The coming of quantum computing and its security implications

42:04 - Quantum threats to cryptography and data security

48:33 - The distinction between quantum-safe cryptography and other security measures

53:08 - The role of industry and capitalism in solving security problems

01:00:07 - Closing remarks and upcoming episodes

Sound Bites:

"Models operating with higher autonomy pose new security challenges."

"AI makes security harder but also easier if the architecture is right."

"Quantum computing will lay waste to classical cryptography."