As enterprises transition from AI experimentation to broad deployment, many are facing a critical and preventable failure: treating security as an afterthought. As AI agents proliferate across business workflows, gaps in identity, governance, and monitoring are creating new attack surfaces that hackers are already exploiting. In this exclusive interview from RSAC 2026, Sunil Agrawal (CISO, Glean) and Michael Sikorski (CTO & VP of Engineering, Unit 42, Palo Alto Networks) join Tom Field to discuss the emerging risks of AI agents and how to secure them at scale. Key topics in this interview: -The New Attack Vectors: Why prompt injection, jailbreaking, and data poisoning are becoming the primary threats to AI-driven workflows. -Foundations of Safe AI: Why identity management and declared guardrails are essential before deploying AI agents. -Glean + Palo Alto Networks: How this integration enforces least-privilege data access, runtime security, and incident response readiness. -Threat Modeling Tools: Resources for security leaders, the latest Unit 42 incident response report. To learn more, visit the Palo Alto Networks partner directory: https://technologypartners.paloaltonetworks.com/English/listing/glean

 162      2