Zscaler has introduced new products and platform updates aimed at securing AI agents. The launch centres on what it describes as a complete zero-trust platform for agentic AI.
The announcement extends Zscaler's Zero Trust Exchange to cover how AI agents connect to systems, access data and operate on devices. It also adds tools to monitor agent activity, manage access controls and identify threats linked to AI software running on employee endpoints.
The move reflects a broader shift in enterprise security as companies deploy autonomous software agents to carry out tasks once handled by staff. These agents can act on behalf of users or independently, and they may create short-lived identities, spawn sub-agents and make access requests at a pace older security products struggle to track.
Zscaler's update is built around two new offerings. One is AI Broker, designed to secure communications involving AI agents through MCP and A2A brokers. The other is Endpoint AI Security, intended to detect and block AI-related threats on user devices, including risks associated with browsers, plugins, extensions and local AI tools.
Zscaler also introduced AI Access Graph, a mapping layer for data and identity connections across an organisation. The technology comes from its acquisition of Symmetry Systems and is now being integrated with the Zero Trust Exchange platform.
The graph is designed to show how users, agents, applications, models and data sources interact across enterprise environments. The goal is to give security teams a clearer view of data lineage and access relationships so they can apply tighter policies and reduce unnecessary access.
Broader update
Zscaler also outlined a broader expansion of AI Protect, a product line launched earlier this year. The latest additions cover three areas: AI asset management, controlled access to approved AI tools, and protection for AI applications during development and runtime.
For asset management, it now offers discovery of embedded AI in software-as-a-service and internet traffic, identification of AI agents and MCP servers in public cloud environments, scanning of codebases for agentic risk, and visibility into AI activity on endpoints.
For access controls, Zscaler has expanded prompt extraction across more than 250 generative AI applications. It also added full conversational views, support for Anthropic and OpenAI compliance APIs, and guardrails designed for multi-turn interactions.
For AI infrastructure and applications, Zscaler added red teaming for MCP servers, a standalone prompt-hardening service and compliance heat maps. These features are intended to help companies govern AI applications more closely as they move from development into live use.
Zscaler framed the launch as a response to changing enterprise security models. Traditional systems were built around known human users and relatively stable access patterns, while AI agents can operate continuously, chain tasks together and interact with a wider range of systems in less predictable ways.
That creates challenges not only for access management but also for tracing where data moves and which identities are responsible for particular actions. Endpoint exposure is also becoming a larger issue as AI tools, browser extensions and plugins create new routes for malicious activity.
Jay Chaudhry, Chairman and Chief Executive Officer at Zscaler, said the new products are intended to address that gap.
"Traditional security was never designed for millions of autonomous agents that act and reach sensitive data at machine speed," said Jay Chaudhry, Chairman and Chief Executive Officer of Zscaler.
"We pioneered Zero Trust Exchange to secure users, branches and cloud workloads, and now we are extending zero-trust security to AI agents. Enterprises are no longer held back from rolling out agents everywhere," Chaudhry said.
The launch also drew support from John Israel, Global Chief Information Security Officer at KPMG, who spoke about the need for better oversight as businesses expand their use of AI agents in operations.
"Managing data security is no longer just about building high walls; it is about scaling visibility and treating data as a highly active, strategic asset," said John Israel, Global Chief Information Security Officer at KPMG, who joined Zscaler as a guest speaker to discuss the launch.
"As businesses scale their use of AI agents to optimize operations, having a unified, zero-trust framework to trace data lineage and govern agent-to-agent interactions is paramount to maintaining trust, compliance and competitive advantage," Israel said.
Zscaler says its platform operates across more than 160 data centres globally and is used by large businesses, critical infrastructure operators and government agencies. The latest additions are intended to help customers apply policy controls across both cloud environments and endpoint devices while keeping closer watch on how AI agents interact with data.