Control 2.9: Defender for Cloud Apps — Copilot Session Controls — Portal Walkthrough
Step-by-step portal configuration for deploying Microsoft Defender for Cloud Apps session controls to monitor and govern browser-based Microsoft 365 sessions used alongside Copilot.
Prerequisites
- Microsoft Defender for Cloud Apps license for each protected user
- Security Administrator or Cloud App Security Administrator role
- Conditional Access integration configured
- Understanding of session control capabilities
Steps
Step 1: Enable Conditional Access App Control
Portal: Microsoft Defender portal Path: Settings > Cloud Apps > Connected apps > Conditional Access App Control apps
Enable Conditional Access App Control for the Microsoft 365 web apps your Copilot users rely on. This allows Defender for Cloud Apps to proxy supported browser sessions and apply real-time controls.
Verify Microsoft 365 appears in the connected apps list with "Enabled" status.
Step 2: Create Session Policy for Copilot Monitoring
Portal: Microsoft Defender for Cloud Apps Path: Defender > Policies > Policy Management > Create Policy > Session Policy
Create a session policy targeting the browser-based Microsoft 365 activity you need to govern: - Policy name: "FSI Copilot Session Validation" for a first-pass routing check, then a separate "FSI Copilot Sensitive Session Audit" policy for detailed monitoring - Session control type: Start with Monitor only to validate redirection, then use Block activities or Control file download/upload (with inspection) for detailed monitoring - Filters: Select the onboarded Microsoft 365 web apps in scope and, where needed, activity types such as printing, clipboard actions, sending, sharing, or editing in supported apps - Actions: Use Audit first for detailed policies; add Block, Protect, or Require step-up authentication only after validation
Step 3: Configure Content Inspection for Copilot Sessions
Portal: Microsoft Defender for Cloud Apps Path: Defender > Policies > [Session Policy] > Content Inspection
Enable content inspection on the detailed session policies to detect sensitive data in supported browser-session actions: - Enable DLP or malware inspection for the selected session control type - Select sensitive information types, sensitivity labels, or classifier conditions relevant to FSI (for example, SSNs, account numbers, or confidential labels) - Configure actions appropriate to the policy type: Audit, Block, Protect, or Require step-up authentication
Step 4: Set Up Real-Time Alerts
Portal: Microsoft Defender for Cloud Apps Path: Defender > Policies > Alert Policies
Configure alerts for the routed web-session events and anomaly detections that matter most: - Sensitive content detected during supported file download, upload, or send/share actions - Policy violation attempts in routed browser sessions - Impossible travel or activity from infrequent country/region after the anomaly-detection learning period - Malware detection for supported file transfers after you explicitly enable the built-in malware policy, which Microsoft documents as disabled by default
Step 5: Configure Activity Logging
Portal: Microsoft Defender portal Path: Defender portal > Cloud Apps > Investigate > Activity Log
Verify that redirected sign-ins and monitored browser-session actions appear in the policy report or activity log. Configure log retention and export settings for compliance documentation. Use Microsoft Purview audit for prompt/response-level Copilot interaction evidence rather than expecting MDCA to expose per-prompt telemetry.
Step 6: Review Generative AI App Catalog
Portal: Microsoft Defender portal Path: Cloud apps > Cloud app catalog, then Cloud apps > Cloud discovery > Discovered apps
- Navigate to the Cloud app catalog and select the Generative AI category filter to display the 1,000+ generative AI apps cataloged by Microsoft
- Review the cloud app catalog risk scores and note any apps that warrant closer governance review
- Open Cloud discovery > Discovered apps and filter by Generative AI to identify the AI apps actually observed in your environment
- Assess risk scores for any generative AI apps employees are using — high-risk apps may represent Shadow AI usage that exposes customer or financial data
- For prohibited apps: mark them as Unsanctioned and use the supported governance stream (such as Microsoft Defender for Endpoint) or a generated block script to enforce blocking
- For approved apps: mark them as Sanctioned if that tag is part of your governance model, and document the review cadence
Recommended governance workflow: - Monthly: Review newly discovered generative AI apps and assess risk scores - Quarterly: Update the sanctioned/unsanctioned app list and review block policies
Step 7: Configure Agent Threat Detection
Portal: Microsoft Defender portal (security.microsoft.com) Path: Defender portal > Settings > Security for AI > Get started; agent inventory at Assets > AI agents
- Confirm the tenant holds a Microsoft Agent 365-eligible license using the Control 2.9 qualifying-license guidance; Microsoft 365 E5 is not an exclusive prerequisite. Complete Agent 365 onboarding — required as of July 1, 2026 for Copilot Studio and Microsoft Foundry agent threat-detection coverage
- On the Security for AI > Get started page, confirm the Security for AI Agents toggle is enabled. For Copilot Studio real-time protection, copy the onboarding URL from Defender and send it to the Power Platform administrator; after they complete the Power Platform configuration, obtain the App ID from them, paste it into Defender, and save the connection.
- Verify that your Copilot Studio agents and any Microsoft 365 agents surfaced by Defender inventory appear in Assets > AI agents with a current risk level
- Navigate to Incidents & alerts and filter for agent-related incidents to confirm detection is operational
- Under Settings > Security for AI > Policies & rules > Real-time protection, review the built-in Default audit rule and create agent-scoped custom rules for documented detection types that require blocking. Use Advanced Hunting telemetry to create custom detections and downstream automation for organization-specific anomalous behavior.
- Integrate agent threat alerts into your SIEM or Microsoft Sentinel workspace for correlation with user activity
FSI Recommendations
| Tier | Recommendation |
|---|---|
| Baseline | Validate browser-session redirection, configure basic alerting, review the generative AI catalog monthly, and enable agent monitoring alerts in Microsoft Defender for supported agent deployments (requires Microsoft Agent 365 license where applicable) |
| Recommended | Add content inspection with DLP or malware integration, review Cloud Discovery generative AI usage, and configure custom agent anomaly detection or hunting rules |
| Regulated | Full supported browser-session control with blocking or step-up authentication, SIEM integration for MDCA and Purview evidence, agent threat detection in SOC playbooks, and quarterly Shadow AI governance review with sanctioned/unsanctioned app list |
Next Steps
- Proceed to PowerShell Setup for session control automation
- See Verification & Testing to validate session controls
- Review Control 2.3 for Conditional Access integration
- Back to Control 2.9