Sam Curry, CISO of Zscaler, discusses the security implications of agentic AI workflows at RSAC 2026 Conference with us. We dig into the implications of AI agents on identity management, authentication frameworks, and security operations. Curry emphasizes the critical need for proper governance, zero trust architecture, and accountability as AI agents become the dominant form of interaction on the internet. But where do you start?

Key takeaways:
• Identity and runtime are the two critical security pillars for agentic AI
• AI agents will soon outnumber human interactions online
• Authentication, authorization, and authenticity verification are essential for agent security
• API security remains a fundamental challenge in the agentic era
• Zero trust principles help contain risk and manage agent sprawl
• Automation must account for intelligent adversaries to avoid predictability
• The future security workforce will blend human expertise with agent assistance

Chapters:
0:47 - Identity and runtime as security fundamentals
1:44 - Silicon-based intelligence and agent interactions
3:11 - Building authentication and authorization frameworks
9:51 - Zero trust architecture and governance
11:53 - API security and MCP protocol challenges
18:45 - Automation and adversarial security challenges
19:57 - The future of human-agent collaboration in security