Palo Alto Networks EVP: Securing AI Agents in the Enterprise
CXOTalk, Monday, July 20th, 2026
Palo Alto Networks' Anand Oswal explains how to secure AI agents across enterprise, SaaS, endpoint and browser surfaces.
In CXOTalk episode 924, Anand Oswal, EVP of Network Security at Palo Alto Networks, tells host Michael Krigsman that enterprises run more AI agents than their security teams realize. Discovery comes first, because you cannot secure agents, models, tools and plugins you cannot see.
Palo Alto Networks found one third of public MCP servers carry takeover-level vulnerabilities, vibe coding agents demand privileged access to local files, terminals and cloud credentials, and browser agents inherit sessions and cookies enabling identity impersonation.
Runtime threats include prompt injection, memory poisoning, tool misuse and model DoS. Oswal argues a centralized AI gateway anchoring identity, runtime and observability beats a stack of point products.