Zscaler Inc. today unveiled a set of products designed to secure autonomous artificial intelligence agents, with the cybersecurity company claiming it has built the industry’s first complete zero-trust platform for agentic AI.

Announced at the company’s Zenith Live 2026 conference in Las Vegas, the new tools extend Zscaler’s Zero Trust Exchange platform to cover how AI agents connect to networks, reach enterprise data and run on employee devices. They address a security model that Zscaler argues is breaking down as companies move from human users to software agents that act on their own.

Autonomous agents operate at machine speed, spin up temporary identities, spawn sub-agents and exercise permissions in ways that tools built around known human users struggle to see or control, the company said. That creates gaps in visibility and governance that obscure agent risk and make data flows hard to track.

At the center of the launch are two products. Zscaler AI Broker secures agent-to-agent and Model Context Protocol communications and ships with an agent registry that lets organizations define what each agent is permitted to access. Zscaler Endpoint AI Security targets AI threats on employee devices, reaching into browsers, extensions, plugins and local AI tools that the company said legacy endpoint products were not built to inspect.

Zscaler also introduced AI Access Graph, a tool that maps how identities, applications and data sources connect across an organization so security teams can spot and cut unnecessary access. The technology comes from Zscaler’s acquisition of Symmetry Systems Inc., a deal announced in May for $175 million. Symmetry built its business around mapping data access across human and nonhuman identities.

The announcements build on Zscaler AI Protect, which launched in January. The company is adding the ability to discover embedded AI in software-as-a-service and internet traffic, identify agents and MCP servers running in public cloud and scan agentic codebases for risk. It’s also expanding controls for sanctioned AI tools, with prompt extraction across more than 250 generative AI apps and support for the compliance application programming interfaces offered by Anthropic PBC and OpenAI Group PBC.

For teams building AI applications, Zscaler is adding red teaming for MCP servers, a standalone prompt hardening service and compliance heat maps meant to strengthen governance across the development and runtime lifecycle.

The push reflects a wider scramble among security vendors to wrap controls around agentic AI before enterprises deploy it at scale. Rivals including Palo Alto Networks Inc. and Google LLC have rolled out their own agent security offerings over the past year and Zscaler has spent recent months acquiring its way into the category, picking up browser security firm SquareX Ltd. in February.

“Traditional security was never designed for millions of autonomous agents that act and reach sensitive data at machine speed,” said founder and Chief Executive Jay Chaudhry. “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.”

Zscaler did not disclose pricing or general availability dates for the new products.

Image: Zscaler