Announcements at Zenith Live

Some of the most compelling announcements from the event included:

• AI Access Graph. While control is ultimately important in any area of cybersecurity, visibility is always the first step. Zscaler introduced AI Access Graph, powered by technology from its Symmetry Systems acquisition. AI Access Graph builds on Zscaler's AI Asset Management capabilities and maps how identities, applications, agents and data connect to AI services across an enterprise, helping security teams accurately assess risk. This visibility will become the intelligence behind enforcement policies.
• AI Broker. Speaking of enforcement, AI Broker was unveiled to secure agentic traffic through both MCP and A2A brokers. An integrated Agent Registry enables security teams to apply fine-grained controls, limiting what AI entities can access across the enterprise.
• Endpoint AI Security. While Zscaler has traditionally focused on the network, it also recognizes the need for endpoint visibility. Endpoint AI Security discovers and controls AI on employee devices across browsers, extensions, plugins and local AI tools.
• Enterprise Browser. Zscaler announced its plans to release a dedicated enterprise browser. The vendor already offers remote browser isolation through Zero Trust Cloud Browser Isolation and an extension via its Zero Trust Browser Extension, which the company believes addresses most customer use cases. However, for situations that require stronger security and more granular control, Zero Trust Enterprise Browser enables customers to choose the option best suited for their needs.
• ZAgent Framework. Zscaler also covered the AI for security side of the equation with its ZAgent Framework. This will serve as the foundation for autonomous SASE, enabling administrators to interact via natural language prompts in the Zscaler Experience Center. The ZAgent orchestrator interprets the intent and routes it to the right specialized agent for the part of the platform in question -- i.e. ZIA, ZPA, SecOps, data security, etc. Ultimately, it will become the operations layer for a headless system.

Putting zero trust at the root of AI security

Many security teams are still wrapping their heads around the challenges AI poses in their environments. While securing AI requires a number of capabilities, the overall strategy should be rooted in zero trust. At its core, zero trust is about shifting security from a static, binary, perimeter-based model to a dynamic, risk-based, distributed architecture. The nondeterministic nature of AI means that simple allow/block policies do not work. It is critical to assess intent, quantify risk and dynamically adapt policies in real time. There's still work to be done to reach this point, but zero trust has prepared security teams to think differently about protecting their environments.

Zscaler has spent years building and refining its platform to apply these types of zero-trust tenets across users, workloads, branches and devices. The announcements at Zenith Live laid out a path to expand the platform and extend those capabilities to generative and agentic AI.

For customers who have already invested in and deployed Zscaler, it's a compelling value proposition. While many enterprises will still hedge their bets and deploy a mix of vendors in the early stages, if Zscaler can deliver on the vision it outlined, it will be well-positioned to lead in zero-trust security for AI.

John Grady is a principal analyst at Omdia who covers network security. Grady has more than 15 years of IT vendor and analyst experience.

Omdia is a division of Informa TechTarget. Its analysts have business relationships with technology vendors.