Zero Trust meets AI: Inside Zscaler’s Zenith Live 2026

For most of the past decade, zero trust was synonymous with one thing: controlling who gets in. At Zenith Live 2026, Zscaler made clear that era is over.

The company’s annual cybersecurity conference, held in Las Vegas, served as the launch pad for what Zscaler is calling the industry’s first complete zero trust platform for agentic AI. This was a significant evolution beyond the secure access function its Zero Trust Exchange was originally built for.

The headline announcements spanned two days of keynotes. 

Day one focused squarely on AI threats. Zscaler unveiled AI Broker for Agents, which enforces access policies between autonomous AI agents over MCP and A2A communications. Alongside it came AI Endpoint Security, targeting threats buried in browsers, extensions, and local AI tools that traditional EDRs simply weren’t built to catch. Rounding out the AI-native additions was AI Access Graph, powered by the Symmetry Systems acquisition, giving security teams real-time visibility into how users, agents, and data sources connect across the enterprise. Rounding out day one was an expansion of AI Protect and the Project AI-Guardian partner ecosystem, drawing in AWS, OpenAI, Google Cloud, and others.

From access control to AI defense mesh

Day two widened the aperture further into SASE territory. Zscaler announced the ZAgent agentic framework to automate SASE administration, a Zero Trust Browser Extension for unmanaged and BYOD devices, Zero Trust B2B Connectivity, and Zero Trust SASE for Multi-Cloud environments including GCP and Kubernetes workloads. 

Also on the table: an Agentic Security Operations Platform that uses Zscaler telemetry to prioritize and respond to threats faster.

The strategic thread running through all of it is straightforward: AI agents don’t behave like human users. They carry ephemeral identities, escalate privileges through chained actions, and operate at machine speed, a reality we explored in depth when we wrote about Zscaler’s Symmetry Systems acquisition. As outlined in our article on treating AI agents like you treat your colleagues, only 21% of organisations currently have full visibility into what their AI agents are accessing, making governance a critical gap.

Zscaler CEO Jay Chaudhry framed the company’s position plainly: “AI agents are able to do the same kind of stuff that your people did. Zscaler is securing users — with similar technology and some additions, we’ll secure agents as well.” 

That’s a confident claim, and the product portfolio announced at Zenith Live 2026 backs it up. But as we have argued, the broader industry still needs to stop just talking about zero trust and start operationalising it, because in the agentic era, the cost of the gap between strategy and execution is no longer theoretical.