Control 3.2: Data Retention Policies for Copilot Interactions — Verification & Testing
Test cases and evidence collection procedures to confirm data retention policies are correctly applied to Copilot interaction data and generated content.
Regulatory applicability
Regulatory citations identify commonly referenced requirements. Applicability, record classification, and retention periods are a firm and counsel determination; this framework does not assert that any regulation applies or is satisfied. See the Regulatory Applicability Questions appendix.
Test Cases
Test 1: Retention Policy Deployment Verification
- Objective: Confirm retention policies are deployed and active for Copilot locations
- Steps:
- Run
Get-RetentionCompliancePolicy | Where-Object { $_.Name -like "*Copilot*" }. - Verify each policy shows
Enabled: TrueandDistributionStatus: Success. - Confirm the Microsoft Copilot experiences policy exists in the Purview portal; current retention policy PowerShell syntax does not expose a Copilot-specific location switch for creating that policy.
- Confirm
TeamsChannelLocationandTeamsChatLocationare set for Teams retention policies. - Expected Result: All FSI Copilot retention policies are deployed and distributed successfully, including the Microsoft Copilot experiences policy.
- Evidence: PowerShell output showing policy status and distribution.
Test 2: Retention Rule Configuration Validation
- Objective: Verify retention duration and actions are correctly configured
- Steps:
- Run
Get-RetentionComplianceRule -Policy "FSI-Copilot-Experiences-Retention". - Confirm
RetentionDurationis 1095 days (3 years) for communications or 2190 days (6 years) for regulated tier. - Confirm
RetentionComplianceActionisKeep. - Expected Result: Retention rules match the FSI-required retention period with retain-only action.
- Evidence: PowerShell output of retention rule configuration.
Test 3: Microsoft Copilot Experiences Location Coverage
- Objective: Validate that Copilot Chat history and meeting recap content is retained by the Copilot experiences policy
- Steps:
- Have a test user perform a Copilot interaction in Microsoft Copilot Chat.
- Wait for retention policy processing (up to 7 days for initial deployment).
- Use Content Search to locate the interaction data in the Copilot experiences location.
- Verify the content is marked as retained per the applied policy.
- Expected Result: Copilot interaction data in the Microsoft Copilot experiences location is discoverable and marked with retention metadata.
- Evidence: Content search results showing retained Copilot Chat interaction data.
Test 4: Retention Label Application
- Objective: Confirm retention labels can be applied to Copilot-generated documents
- Steps:
- Create a document using Copilot assistance in Word or Excel.
- Apply the
FSI-Copilot-Regulatory-Record-6yrlabel manually or via auto-labeling. - Verify the label metadata appears in the document properties.
- Expected Result: Retention label is applied and the document shows the correct retention period.
- Evidence: Screenshot of document properties showing applied retention label.
Test 5: Threaded Summary Retention Behavior
- Objective: Confirm that Copilot-generated meeting summaries are retained independently from source content deletion
- Steps:
- In a test Teams meeting, use Copilot to generate a meeting recap.
- Verify the meeting recap appears in the Teams channel or meeting chat.
- Delete the Teams meeting transcript (or a specific Teams message) and wait 24 hours.
- Confirm that the Copilot-generated meeting recap (threaded summary) remains accessible — it should be retained independently by the Microsoft Copilot experiences retention policy.
- Use Content Search in Purview to verify the summary is still discoverable after source content deletion.
- Expected Result: The Copilot-generated meeting recap remains retained and discoverable after the source Teams meeting content is deleted. Firms should ask counsel to review whether the evidence is appropriate for FINRA Rule 4511(c) preservation expectations for source and summary content.
- Evidence: Content search results showing the Copilot summary is retained independently of the deleted source content; screenshots comparing content before and after source deletion.
Test 6: New Retention Location Categories Validation
- Objective: Verify that the Purview portal shows the restructured retention location categories and that the correct location is selected for M365 Copilot
- Steps:
- Navigate to Microsoft Purview portal > Data lifecycle management > Retention policies.
- Click New retention policy (or open an existing policy) and review the available location options.
- Confirm that Microsoft Copilot experiences, Enterprise AI Apps, and Other AI Apps are available as distinct location categories.
- Verify that the FSI retention policies use Microsoft Copilot experiences (not Enterprise AI Apps or Other AI Apps).
- Expected Result: The three Purview retention location categories are visible. All FSI M365 Copilot retention policies target the Microsoft Copilot experiences location.
- Evidence: Screenshot of retention policy location selector showing the three categories; screenshot of existing policies confirming Microsoft Copilot experiences is selected.
Test 7: Priority Cleanup Approval Verification (Recommended Tier — if configured)
- Objective: Confirm that Priority cleanup is used only for a documented Microsoft-supported storage-remediation scenario and follows the required simulation and approval workflow
- Steps:
- Review the Priority cleanup request and confirm the scenario is documented (for example, stale Teams meeting recordings/transcripts or departed-user Preservation Hold Library cleanup).
- Verify the mandatory simulation completed and the reviewed item sample matches the approved scenario.
- Confirm that records and regulatory records are excluded from scope.
- Confirm required approvals are recorded, including eDiscovery Administrator approval if any eDiscovery hold applies.
- Verify the cleanup rationale is documented in the firm's records management schedule.
- Expected Result: Priority cleanup is narrow in scope, simulation-backed, approved by the required roles, and does not target records or regulatory records.
- Evidence: Priority cleanup simulation report, approval record, exclusion evidence for records/regulatory records, and records management schedule entry showing regulatory rationale.
Evidence Collection
| Evidence Item | Source | Format | Retention |
|---|---|---|---|
| Retention policy configuration | PowerShell | Text export | With control documentation |
| Retention rule details | PowerShell | Text export | With control documentation |
| Microsoft Copilot experiences policy distribution | PowerShell/Portal | Screenshot | With control documentation |
| Threaded summary retention test | Purview Content Search | CSV/Screenshot | Per policy |
| Label application proof | SharePoint/OneDrive | Screenshot | With control documentation |
| Location category verification | Purview portal | Screenshot | With control documentation |
Compliance Mapping
| Regulation | Requirement | How This Control Helps |
|---|---|---|
| SEC Rule 17a-4(a) | 6-year broker-dealer-record citation | Microsoft Copilot experiences retention policy can be configured for a 6-year firm-approved example; applicability: confirm |
| SEC Rule 17a-3(a)(17) | Communications-record citation | Conservative Regulated-tier retention of all Copilot-generated content is a firm-approved example pending counsel confirmation |
| FINRA Rule 4511 | Books-and-records retention citation | AI interaction records in scope across Copilot surfaces: confirm |
| FINRA Rule 4511(c) | Preservation format and media citation | Threaded summary retention test provides evidence for counsel review of source and summary preservation |
| GLBA | Financial-record/customer-information citation | Privacy and record-preservation applicability for AI-processed customer information: confirm |
Next Steps
- Review Troubleshooting for retention policy issues
- Proceed to Control 3.3 for eDiscovery configuration
- Back to Control 3.2