Protect sensitive AI prompts and files with Microsoft Purview network DLP
Implementation Effort: Medium - Requires joint network and data-protection ownership of acquisition, inspection, policies, and monitoring.
User Impact: Medium - Sensitive content can be blocked; inspection limits and false positives need operational handling.
Overview
Global Secure Access (GSA) content policies can use Scan with Purview to inspect supported text and files for Microsoft Purview data loss prevention (DLP) decisions. This extends protection beyond destination blocking to sensitive content exchanged with unmanaged or cloud AI applications.
This supports Use least privilege access for data movement and Assume breach by adding inline controls on eligible traffic. Basic content filtering, collection-only policies, and Purview-backed enforcement remain different capabilities.
Prerequisites
- Complete AI network readiness, including acquisition and Transport Layer Security (TLS) inspection.
- Appropriate GSA/Purview administrative roles and licensing. The supplied licensing reference lists Microsoft 365 E7 or Purview E5/equivalent plus Entra Internet Access/equivalent for this integration.
- Traffic mapped to an Entra user identity.
- Matching GSA content-policy and Purview DLP scope, joint operational ownership, a pilot scope, and approved synthetic test data.