Why Southeast Asia CISOs Need Zero Trust as Their AI Control Plane – AI Agents, Data Borders and Supply Chains

At Zenith Live 2026 held on 16-17 June in Vienna, Zscaler sharpened a reality that Southeast Asia CIOs and CISOs are already sensing, which are, AI agents are quickly becoming digital workers inside their organisations, while regulators tighten data residency rules and supply‑chain attacks move closer to core business operations.

Zscaler’s solution is to extend its Zero Trust Exchange and SASE platform beyond users and workloads to AI agents, unmanaged devices, multi‑cloud workloads, and B2B partners, effectively positioning zero trust as the control plane for secure AI adoption in highly connected, highly regulated markets like Southeast Asia.

In my opinion, three moves stand out for Southeast Asia organisations at the AI layer:
1. An AI Broker with an Agent Registry that governs how AI agents talk to data, applications, and other agents, inspecting prompts and responses and enforcing least‑privilege access in real time. In my view, this is critical in sectors facing strict data‑handling rules across multiple jurisdictions.
2. Endpoint AI Security that exposes risky local AI tools, browser extensions, and plugins proliferating on endpoints across distributed workforces and contractor ecosystems common in Southeast Asia.
3. An AI Access Graph and AI Protect that map AI assets, model usage, and data flows across SaaS, public cloud, and on‑prem, backed by red‑teaming, prompt hardening, and guardrails for more than 250 GenAI apps.

Equally important for Southeast Asia region is how Zscaler handles cross‑border connectivity and sovereignty. The company’s Zero Trust B2B Exchange replaces site‑to‑site VPNs and MPLS links with policy‑controlled application access, so partners, outsourcers, and regional subsidiaries never sit on the same network. This is even as data and workflows move between markets. In parallel, its cloud is engineered for strict locality of logs and operations, with regional data centres and no external “kill switches”, a design clearly influenced by European GDPR and localisation demands that now echo in Southeast Asian data regimes.

On the ground, customer stories from AkzoNobel and Siemens Healthineers show what this looks like when applied decisively – “dark” branches that cannot be discovered on the internet, zero‑trust based B2B connectivity, and an explicit strategy to guide AI adoption rather than banning it.

For Southeast Asia CIOs and CISOs, the practical message is clear:
1. Build a live inventory of AI usage and data flows across borders before regulators and auditors force the issue.
2. Hide your infrastructure and supply chain behind zero trust, so neither partners nor AI agents can turn a single misconfiguration into a regional incident.
3. Treat zero trust as your AI operating model, not a side project, because every new AI agent you deploy is now part of your workforce, your compliance posture, and your attack surface.

My Recommendations: 3 Immediate Priorities for Southeast Asian CISOs in the AI Era
1. Reframe the Threat Model Around Agents, Not Just Users  
a. Update threat models and control frameworks to explicitly include AI agents as identities: what they can access, what actions they can perform, and how they are monitored.
b. Classify agents by criticality and blast radius in the same way you do privilege human accounts and critical applications.

2. Cut Lateral Movement Before You Chase Every Vulnerability 
a. Assume you will never patch everything, focus first on eliminating discoverability and lateral movement across branches, factories, and multi‑cloud workloads.
b. Use zero trust segmentation so a compromised agent, endpoint, or partner connection can only see and touch what policy explicitly allows.

3. Operationalise AI Guardrails and Evidence for Regulators 
a. Implement AI‑aware controls: AI Broker, guardrails for GenAI apps, data lineage via access graphs, and endpoint visibility into AI tools.
b. Ensure you can produce evidence such as logs, policies, lineage, showing how AI access is governed across borders, partners, and regulated datasets.