2.4.4. Reviewing and Responding to DLP Alerts
💡 First Principle: A DLP match is one event; an alert is that event judged worth a human's attention. Because the same user is often also the subject of phishing or endpoint alerts, Microsoft routes DLP alerts into Defender XDR, where all of them are correlated into one incident.
Where to investigate. Purview creates the policies; Microsoft recommends the Defender portal for investigating DLP alerts. In Defender XDR, DLP alerts are correlated into incidents with alerts from Defender for Office 365 and Defender for Endpoint, and you can act on SharePoint and OneDrive files directly (apply a label, remove sharing, delete). Find them under Incidents & alerts by filtering detection source to Microsoft Data Loss Prevention.
| Purview DLP alerts dashboard | Defender XDR | |
|---|---|---|
| Alert retention | 30 days | 6 months |
| Correlation with other threats | No | Yes — incidents span DLP, email and endpoint |
| Microsoft's recommendation | Policy authoring | Investigation |
Generating alerts. Each DLP rule's incident-report settings decide whether a match raises an alert, its severity and who's notified; aggregation (alert once per volume or per time window) depends on licensing. DLP integration with Defender XDR requires E5-level licensing (Microsoft 365 E5, Office 365 E5, E5 Compliance or E5 Information Protection & Governance).
Copilot and AI alerts follow the same path: a DLP rule on the Copilot location can generate alerts, and the DSPM activity explorer shows the AI interactions behind them.
A recent change: since early October 2026, a built-in Defender alert tuning rule treats DLP signals as behaviors, so they don't appear as alerts in the incident queue unless you disable that rule under Settings > Microsoft Defender XDR > Alert tuning.
⚠️ Exam Trap: If an investigator needs to review a DLP alert from three months ago, the Purview dashboard can't help — it keeps alerts for 30 days. Defender XDR keeps them for six months.
Reflection Question: An analyst sees a DLP alert for a user emailing customer records externally, and also a phishing alert for the same user the day before. Why is Defender XDR the better place to investigate, and what would you look for?