Skip to content

Control 1.5: Data Loss Prevention (DLP) and Sensitivity Labels - Troubleshooting

This playbook provides troubleshooting guidance for Control 1.5.

Common Issues

Issue: DLP Policy Not Detecting Sensitive Content

Symptoms: Sensitive data flows through AI without triggering DLP alerts

Solutions:

  1. Verify policy is in "Enable" mode (not "Test" mode)
  2. Confirm policy locations include AI locations (Microsoft 365 Copilot / Copilot Studio)
  3. Confirm SIT readiness and scope (see Control 1.13):
  4. The SIT exists and is enabled
  5. The SIT pattern actually matches your test data
  6. Minimum count/confidence thresholds are not too strict
  7. Confirm you are testing a supported content path:
  8. For label-based rules, confirm the label is applied to the item (file/message), not only a container
  9. For endpoint tests, confirm Devices/Endpoint DLP is enabled and targeted
  10. Allow for propagation time before re-testing (often hours; in some cases longer)

Issue: False Negatives in AI Prompts/Responses

Symptoms: Some prompts/responses containing sensitive patterns are not detected

Solutions:

  1. Reduce reliance on a single pattern: add supporting keywords or additional SITs
  2. Validate formatting variations (spaces/hyphens/prefixes) in SIT patterns per Control 1.13
  3. Review rule logic (AND/OR) and priority; ensure an allow rule is not bypassing detection
  4. Confirm the AI location is included (a rule scoped to a different workload will not trigger)

Issue: Too Many False Positive DLP Alerts

Symptoms: Legitimate business content triggers DLP blocks

Solutions:

  1. Review SIT confidence levels and minimum count; increase thresholds where justified
  2. Add corroborating context (supporting keywords) to custom SITs and tighten regex patterns
  3. Add scoped exceptions only when you can justify and evidence them
  4. Prefer "Warn" + audit for ambiguous detections; reserve "Block" for high-confidence content
  5. Re-run the test suite after tuning and retain evidence of before/after results

Issue: Sensitivity Labels Not Enforcing in AI

Symptoms: Content with "Highly Confidential" label still accessible to agents

Solutions:

  1. Verify DLP rule includes label-based conditions
  2. Check label is correctly applied to content (not just container)
  3. Confirm label policy is published to affected users
  4. Verify DSPM for AI integration is enabled
  5. Check if agent service account is in label scope

Issue: DSPM Oversharing Assessment Shows No Results

Symptoms: Assessment completes but shows zero overshared items

Solutions:

  1. Verify data sources are correctly specified
  2. Check that content has sensitivity labels applied
  3. Confirm assessment scope includes the correct sites
  4. Wait for assessment processing (can take 24-48 hours)
  5. Verify permissions to access assessment results

Issue: DLP Blocking Legitimate Agent Operations

Symptoms: Agent cannot access required data due to DLP blocks

Solutions:

  1. Review incident details to understand what triggered block
  2. Create exception for agent service account if appropriate
  3. Adjust SIT minimum count thresholds
  4. Use contextual conditions to allow specific scenarios
  5. Consider separate policy for agent service accounts

Issue: Virtual Connector Classification Not Enforcing

Symptoms: Agent uses blocked connector without triggering DLP error; expected DLP block does not occur

Solutions:

  1. Verify DLP policy scope includes the target environment:
  2. Navigate to PPAC > Policies > Data policies > [Policy] > Environments
  3. Confirm target environment is listed in policy scope
  4. Note: Policies scoped to "All environments except..." may not apply as expected
  5. Check policy propagation timing:
  6. Allow 1-2 hours after policy changes for propagation
  7. Clear browser cache and retry agent creation/publishing
  8. Verify connector is correctly identified:
  9. Some connectors have multiple versions or aliases
  10. Ensure you've blocked all variants (e.g., "HTTP Webhook" vs "HTTP with no authentication")
  11. Check for conflicting policies:
  12. Multiple DLP policies may apply to the same environment
  13. Higher-priority policy may override expected classification
  14. Review all policies via Get-DlpPolicy PowerShell cmdlet
  15. Verify agent is in target environment:
  16. Agents in different environments may have different DLP policies
  17. Confirm agent environment matches policy scope

Issue: HTTP Endpoint Filtering Not Blocking Expected URLs

Symptoms: Agent successfully calls external API that should be blocked by endpoint filtering; allow list not enforcing

Solutions:

  1. Verify endpoint filtering is configured:
  2. Navigate to PPAC > Policies > Data policies > [Policy] > Connectors
  3. Click "HTTP with Microsoft Entra ID" connector
  4. Confirm endpoint filtering badge/icon is present
  5. Click connector to view configured patterns
  6. Check URL pattern syntax:
  7. Patterns are case-sensitive for path components (not domain)
  8. Wildcards (*) must be used correctly: *.domain.com (subdomain wildcard) vs https://domain.com/* (path wildcard)
  9. Verify no typos in domain names
  10. Verify filtering mode:
  11. Confirm you're using "Allow list" mode (not "Block list") for Zone 3
  12. In Allow list mode, only specified patterns are permitted
  13. In Block list mode, only specified patterns are denied
  14. Allow for propagation delay:
  15. Endpoint filtering changes can take 1-2 hours to propagate
  16. Test in incognito/private browsing mode to avoid cache
  17. Check for pattern overlap:
  18. Broader pattern may allow URL you intended to block
  19. Example: *.bank.com allows https://external.bank.com even if you intended only internal.bank.com
  20. Verify agent is using HTTP with Microsoft Entra ID connector:
  21. If agent uses HTTP Webhook (different connector), endpoint filtering for HTTP with Entra ID won't apply
  22. Check agent skills/actions to confirm connector type

Issue: Maker Sees No DLP Error When Using Blocked Connector

Symptoms: Agent maker can add blocked connector to agent without error message; no DLP notification

Solutions:

  1. Verify DLP policy applies to maker's user account:
  2. Check if maker is in excluded group (some policies exclude admins)
  3. DLP policies apply based on environment membership, not user role
  4. Confirm environment-level DLP policy scope:
  5. Navigate to PPAC > Policies > Data policies > [Policy] > Environments
  6. Verify the environment where agent is being created is in policy scope
  7. Default environment may have different policies than custom environments
  8. Check for "Audit only" mode:
  9. If policy is in "Audit only" mode, makers see no blocking errors
  10. Verify policy mode in PPAC (should be "Enabled" for enforcement)
  11. Test with actual agent execution (not just design-time):
  12. DLP enforcement may occur at runtime (when agent executes) rather than design-time (when agent is created)
  13. Publish agent and test execution to verify DLP enforcement
  14. Check connector classification group:
  15. If connector is "Non-Business" (not "Blocked"), maker can use it if not mixing with Business connectors
  16. Verify connector is in "Blocked" group for complete prohibition
  17. Review Purview Audit Log:
  18. Even if maker sees no error, DLP events may be logged
  19. Search Purview Audit Log for DLP policy matches with maker's user account

Escalation Path

If issues persist after troubleshooting:

  1. First tier: Purview Compliance Admin - verify policy configuration
  2. Second tier: Information Protection Team - review label design
  3. Third tier: Microsoft Support - platform-level issues

Back to Control 1.5 | Portal Walkthrough | PowerShell Setup | Verification Testing

Updated: January 2026 | Version: v1.2