Zero Trust for Artificial Intelligence
Zscaler, Friday, October 16th, 2026
Zscaler lays out nine Zero Trust principles for AI systems as agentic AI raises the stakes for least privilege.
Zscaler draws on more than a decade of Zero Trust deployments with ZIA and ZPA, where the goal was least-privilege access for users on trusted devices.
It argues that the next wave of users will include AI agents, which raises the stakes for least privilege.
The post presents nine Zero Trust principles for AI systems, with the intended outcome of AI that is audited, attested and governed.