Zscaler Launches Zero Trust Platform for Agentic AI Security

LAS VEGAS — Zscaler announced a broad expansion of its AI security portfolio, introducing new capabilities designed to secure autonomous AI agents, agent-to-agent communications, and AI-powered applications. Unveiled at Zenith Live 2026, the updates position the company’s Zero Trust Exchange platform as a framework for governing how AI agents access data, interact with systems, and operate across enterprise environments.

The announcement reflects a growing challenge facing enterprise security teams as organizations deploy autonomous AI agents capable of acting independently, spawning sub-agents, accessing enterprise resources, and interacting with other AI systems. Zscaler said traditional security architectures built around human identities and static access controls lack the visibility and governance needed for machine-speed agentic environments. To address these concerns, the company introduced Zscaler AI Broker, which secures communications through Model Context Protocol (MCP) and agent-to-agent (A2A) brokers, while maintaining an integrated Agent Registry that tracks agent permissions and enforces granular access controls.

Zscaler also launched Endpoint AI Security, extending protection to employee devices by detecting AI-related threats embedded in browsers, extensions, plugins, and local AI tools. In addition, the company unveiled AI Access Graph, powered by technology from its acquisition of Symmetry Systems, to map relationships among users, agents, applications, models, and data sources. The platform provides visibility into identity lineage and data flows across enterprise environments, enabling policy enforcement, governance, and risk reduction. The announcement also expands Zscaler AI Protect with new capabilities for AI asset discovery, secure access controls across more than 250 generative AI applications, AI code scanning, AI red teaming for MCP servers, prompt hardening services, and compliance monitoring.

• Introduced Zscaler AI Broker for securing MCP and agent-to-agent communications

• Added integrated Agent Registry for visibility and governance of AI agents

• Launched Endpoint AI Security to protect devices from AI-related threats in browsers, plugins, and local AI tools

• Unveiled AI Access Graph to map relationships among users, agents, applications, models, and enterprise data

• Expanded AI Protect with AI asset discovery, code scanning, conversational monitoring, compliance APIs, AI red teaming, and prompt hardening

• Supports visibility and governance across agentic AI environments, including identity and data lineage tracking

“Traditional security was never designed for millions of autonomous agents that act and reach sensitive data at machine speed,” said Jay Chaudhry, Chairman and CEO of  Zscaler. “We pioneered Zero Trust Exchange to secure users, branches and cloud workloads and now we are innovating to extend the Zero Trust security to AI Agents.”

🌐 Analysis: The launch marks Zscaler’s most comprehensive move yet into AI-native security and follows its January 2026 introduction of AI Protect. The company’s acquisition of  Symmetry Systems has become a key component of its strategy, providing the data lineage and identity graph capabilities required to govern increasingly autonomous AI environments. As enterprises move from AI assistants to autonomous agents, visibility into agent permissions, communications, and data access is emerging as a new security control plane.