Microsoft Defender XDR will integrate AI-generated summaries and categorizations for DLP alerts via the Microsoft Purview Data Security Triage Agent, improving alert triage. Deployment starts April 2026 (preview) and August 2026 (general). Agent management remains in Purview; DLP policies and user impact remain unchanged.
Introduction
We’re introducing Data Security Triage Agent summaries and categorizations for Data Loss Prevention (DLP) alerts directly within the Microsoft Defender XDR portal. This update helps security analysts triage DLP alerts more efficiently by surfacing AI-generated summaries and categorizations created by the Microsoft Purview Data Security Triage Agent.
Screenshot 1: Data Security Triage Agent outputs and summaries now available in DLP alerts in Microsoft Defender XDR
This message is associated with Roadmap ID 558860.
When this will happen:
How this affects your organization:
Who is affected:
What will happen:
Screenshot 2: Security Analysts and Admins triaging DLP alerts in Defenders will be able to deploy the Data Security Triage Agent from the Microsoft Defender XDR portal
What you can do to prepare:
Learn more: Before rollout, we will update this post with new documentation.
Compliance considerations:
| Compliance area | Explanation |
|---|---|
| AI/ML or agent capabilities interacting with customer data | This change introduces AI-generated summaries and categorizations for DLP alerts using the Microsoft Purview Data Security Triage Agent, which processes existing DLP alert data to assist analysts during triage. |
| Admin controls | Admins can deploy the Data Security Triage Agent from the Microsoft Defender XDR portal. Ongoing agent management, including custom instructions, pausing or deactivating the agent, and monitoring usage, remains available in the Microsoft Purview portal. |
| Admin monitoring and compliance reporting | The update enhances DLP alert investigations by adding AI-generated context, improving how admins monitor and assess data security incidents without changing underlying DLP policy enforcement or audit logging. |