en.Wedoany.com Reported - U.S.-based Zscaler has placed AI agent communication governance at the core of its zero-trust security expansion. On May 21, Zscaler announced its intention to acquire U.S.-based Symmetry Systems, planning to integrate the latter's identity mapping and data access graph capabilities into the Zscaler Zero Trust Exchange platform. This will be used to identify, map, and govern communication relationships between AI agents and applications, data, and other agents.
The key focus of this deal is not simply expanding the security product portfolio, but turning the question of "who can access what data, through which systems, and whether an AI agent is involved in the access" into a matter that is visible, auditable, and enforceable by policy. Symmetry Systems' access graph can ingest access logs from enterprise-wide SaaS applications, public cloud services, data storage, and AI systems, and use AI to correlate access relationships between identities, applications, and data. Identities here include both human users like employees and non-human identities such as service accounts, applications, automated processes, and AI agents. Zscaler believes this type of visibility capability is the foundation for enterprises to manage AI agent communications at scale, because security teams must first see the data flows and permission relationships clearly before they can formulate and enforce access boundaries.
Symmetry Systems' technical focus is on placing data context and identity context into the same graph. The company states that its Data Access Graph architecture is designed to help enterprises understand how risk accumulates across identities, data, permissions, and AI systems, and is already used in its DataGuard and AIGuard platforms. For organizations deploying enterprise agents, AI assistants, and RAG applications, the risk comes not only from model outputs but also from whether an agent reads data it shouldn't or transmits sensitive information to the wrong system.
Zscaler plans to combine Symmetry Systems' capabilities with the Zero Trust Exchange to form a data access control layer for the AI era. Traditional zero trust emphasizes least-privilege control for users, devices, applications, and network access. However, as AI agents enter enterprise workflows, the security perimeter expands from "humans accessing systems" to "agents performing tasks on behalf of humans or systems." An enterprise-grade AI agent might simultaneously call upon email, documents, CRM, code repositories, data warehouses, and ticketing systems. Without a cross-system access graph, security teams struggle to determine if each step of its actions complies with data provenance, permission scopes, and business context requirements. An example provided by Symmetry Systems shows that enterprises can define natural language-style information flow policies, such as: no identity within one organizational context, whether human or AI, may read, copy, infer, or transmit data originating from another organization. Enforcing such a policy requires simultaneously knowing what the data is, where it comes from, which identities can reach it, and whether the data flow has crossed boundaries.
This acquisition also reflects a shift in enterprise AI security from model protection to data flow protection. Zscaler emphasizes that the foundational visibility brought by Symmetry Systems will be used to govern communications between AI agents, between AI agents and applications, and between AI agents and data. For industries heavily reliant on data permissions, such as financial services, retail, HR technology, healthcare, telecommunications, manufacturing, energy, and government, the prerequisite for AI agents to truly enter business systems is not simply "connecting to a model," but linking data classification, identity permissions, access paths, policy enforcement, and audit records into a controllable mechanism.
With the planned acquisition of Symmetry Systems, Zscaler's zero-trust security narrative extends further from network and application access to AI agent communication. When enterprises introduce agent workflows, what they need is not just generative capability, but knowing which data the agent accessed, under what identity it performed actions, and whether it touched sensitive boundaries. If Symmetry Systems' access graph capabilities can be fully integrated with Zscaler's existing platform, it will provide a more concrete entry point for visibility and policy enforcement in AI agent security.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
