Global security firm Zscaler has unveiled new products aimed at securing autonomous AI agents, SiliconANGLE reported on June 9.

Zscaler showcased two security products for AI agents and an AI Access Graph at its Zenith Live 2026 event in Las Vegas, Nevada.

The company said it took into account that as enterprise environments shift from being centered on human users to autonomous AI agents, existing security systems make it difficult to control the exercise of privileges and the flow of data.

There are two core products. Zscaler AI Broker protects communications between agents and communications using the Model Context Protocol (MCP). It also manages each agent’s access rights through an agent registry.

Zscaler Endpoint AI Security helps respond to AI threats arising from employee device browsers, extensions, plugins and local AI tools.

Zscaler’s AI Access Graph, based on technology from Symmetry Systems that Zscaler said on May it would acquire for $175 million, identifies connections among identities, applications and data sources within an organization to find and reduce unnecessary access permissions.

Zscaler also strengthened AI Protect, which it launched in January. The update adds features to identify agents running in public clouds and MCP servers, extract prompts targeting more than 250 generative AI apps, and support Anthropic and OpenAI compliance APIs.