Back Issues/Search Home → Calendar → Archive → RSS → Subscribe → Services → Vendors → Blog → Current Issue → Popular → Newsletters (LinkedIn, Substack)

All issues › Volume 344, Issue 2 › IT Vendor News › Varonis

Complete AI Security Starts with the Agent

Varonis, Monday, November 9th, 2026

Varonis strategy VP Ron Bennatan argues AI security should start with the agent itself and run from a single adaptable control plane.

Ron Bennatan, VP of Strategy at Varonis, argues that securing AI one tool at a time leaves teams playing whack-a-mole as agents move across networks, endpoints and applications.

He says complete AI security needs a single control plane that begins with agents: what they can reach, what they need to accomplish and what they are allowed to do.

Effective security requires four unified layers: browser and endpoint protection, prompt and tool governance, intent enforcement and threat detection.

He recounts his team's coding agents unexpectedly building a queue to coordinate access to a shared resource, illustrating that agent behavior cannot be fully anticipated, and notes that unified means shared context applied consistently, not one chokepoint.

more →  ·  More from Varonis →