Zscaler extends Zero Trust to AI for Federal Civilian agencies

Zscaler has set out how Federal Civilian agencies can extend Zero Trust principles to AI, in the final post of a three-part series on AI security and governance. The company said AI introduces new interaction points, including prompts, responses, models, agents, AI APIs, embedded AI and RAG pipelines, each of which can become a path for data exposure or unauthorised access.

Zscaler’s platform covers three areas: AI asset management to discover and inventory AI use across endpoints, traffic, SaaS, code and cloud; secure access to AI applications through access control, DLP, browser isolation and prompt and response inspection; and protection for agency-built AI applications and agents via API-based guardrails, AI red teaming and continuous monitoring.

The vendor framed the work around M-25-21 compliance, saying agencies should start with visibility, then map AI use to data sensitivity and mission risk before enforcing acceptable-use and data protection policies.