Skip to content

Control 2.9: Defender for Cloud Apps — Copilot Session Controls

Control ID: 2.9 Pillar: Security & Protection Regulatory Reference: GLBA §501(b), FFIEC Last Verified: 2026-09-29 Governance Levels: Baseline / Recommended / Regulated


Objective

Deploy Microsoft Defender for Cloud Apps (MDCA) session policies and access policies to provide browser-session governance, documented anomaly detections, and cloud-app oversight for supported Microsoft 365 web sessions used alongside Copilot, while also discovering and governing Shadow AI usage. MDCA extends Conditional Access by enabling granular browser-session controls and connected-app analytics, helping organizations enforce real-time access decisions based on session context. This control supports compliance with GLBA safeguard requirements and FFIEC expectations for continuous monitoring of privileged application access.


Why This Matters for FSI

  • GLBA §501(b) requires monitoring and testing the effectiveness of safeguards — MDCA provides continuous, real-time monitoring of supported browser sessions and cloud-app activity around Copilot-governed data, extending beyond point-in-time authentication controls
  • FFIEC IT Examination Handbook (Information Security) expects continuous monitoring and anomaly detection for critical applications — Copilot's access to enterprise data makes it a high-value monitoring target
  • FFIEC IT Examination Handbook (Audit) expects logging and monitoring of access to sensitive systems — MDCA session logs provide granular visibility into supported browser-session activity and cloud-app events around Copilot-governed data
  • NYDFS Part 500 (Section 500.06) requires an audit trail designed to detect and respond to cybersecurity events — MDCA alerts on anomalous session and cloud-app activity support this requirement
  • Institutions should define approved AI-tool monitoring expectations for safety, soundness, and consumer protection with legal, compliance, and risk owners; MDCA can support that monitoring for Copilot.

Control Description

Microsoft Defender for Cloud Apps integrates with Microsoft Entra Conditional Access to provide browser-session controls that go beyond authentication-time decisions. For Copilot governance, MDCA is most useful as a session-control layer around supported Microsoft 365 web apps and as a Shadow AI discovery/governance surface for generative AI apps. Use Microsoft Purview audit and Copilot security records for prompt/response-level telemetry; MDCA session controls govern the routed browser session, not a dedicated per-prompt Copilot control plane.

MDCA Capability Matrix for Copilot

Capability Description FSI Application
Session policies Real-time monitoring and control for routed browser sessions in supported cloud apps Govern downloads, uploads, clipboard, print, send/share, and step-up authentication actions in Microsoft 365 web sessions used alongside Copilot
Access policies Control access based on client and session context Force browser-only access where session controls must apply and block unsupported native clients
Anomaly detection Built-in UEBA/ML detections over connected apps and sessions Identify impossible travel, infrequent country/region, malware, and similar anomalies relevant to Copilot-adjacent Microsoft 365 activity
Activity policies Alert on specific activity patterns in supported connected apps Notify on risky SharePoint, OneDrive, or Teams web activity that can expose Copilot-grounded data
App governance Monitor OAuth app permissions Track apps that integrate with Copilot
File policies API-connected file monitoring in supported apps Use only while planning migration to Microsoft Purview DLP or auto-labeling because Microsoft retires file policies on January 6, 2027
Generative AI app catalog Inventory and risk assessment of 1,000+ generative AI apps Identify and govern Shadow AI tools used alongside Copilot
Agent threat detection Detection of compromised or malicious agent behavior in Microsoft Defender (requires Microsoft Agent 365 license as of July 1, 2026) Protect Copilot agent deployments from threat actors

Generative AI App Catalog

The Microsoft Defender for Cloud Apps cloud app catalog includes 1,000+ generative AI apps in the generative AI subcategory (part of the overall 31,000+ app catalog). This catalog enables FSI organizations to:

  • Inventory Shadow AI usage: Discover which generative AI applications employees are using without IT authorization alongside sanctioned Copilot deployments
  • Assess risk scores: Each app in the catalog carries a risk score based on security, compliance, and regulatory factors — enabling risk-based governance decisions
  • Apply governance policies: Block or restrict access to high-risk generative AI apps while permitting sanctioned Copilot usage
  • Monitor AI app trends: Track adoption patterns of generative AI tools across the organization to identify emerging Shadow AI risks

Discovery workflow: Microsoft Defender portal > Cloud Apps > Cloud app catalog > Filter: Category = "Generative AI"

FSI relevance: GLBA §501(b) requires safeguards to protect customer information across all technology used to process that data. Employees using unsanctioned generative AI tools (Shadow AI) may inadvertently expose customer information to apps without contractual data protection commitments. The MDCA catalog enables proactive Shadow AI governance.

Agent Threat Detection in Microsoft Defender XDR

Microsoft's November 2025 AI-security release notes identify Microsoft Copilot Studio AI agent protection as Preview. The same release notes list Microsoft Foundry separately as an Agent Control Plane capability; they do not assign the Copilot Studio preview date to Foundry. Microsoft’s transition guidance separately applies the July 1, 2026 licensing cutover to both Copilot Studio and Microsoft Foundry agents. As of July 1, 2026, agent discovery, posture, and threat-detection coverage previously delivered through Microsoft Defender for Cloud Apps or Microsoft Defender for Cloud requires an Agent 365-eligible license. Tenants without an Agent 365-eligible license lose access to those capabilities; equivalent detections continue for licensed tenants through Agent 365 observability logs and the agent registry.

Agent 365 Qualifying-License Guidance

The transition guidance uses the term Agent 365-eligible license. At least one user must hold a qualifying Microsoft Agent 365 license to enable Agent 365. Microsoft 365 E5 is a documented enterprise path, but it is not an exclusive prerequisite. Apply the current Microsoft licensing guidance for the customer's purchasing channel and verify the tenant's entitlement before relying on agent-security coverage:

Customer context Current documented qualifying path
Enterprise Microsoft 365 E5; or Microsoft 365 E3; or Office 365 E3 plus EMS E3, with both Defender Suite and Purview Suite for the E3-based paths
Microsoft 365 E7 Agent 365 is included with Microsoft 365 E7
Frontline Microsoft 365 F1/F3 plus Defender Suite for frontline workers and Purview Suite for frontline workers
SMB Microsoft 365 Business Premium; Defender Suite for SMB and Purview Suite for SMB unlock Agent 365 functionality but are not formal prerequisites
Education Microsoft 365 A5 for Faculty; or Microsoft 365 A3 (or Office 365 A3 plus EMS A3) for Faculty, with both Defender Suite for Education and Purview Suite for Education

This is a licensing-eligibility summary, not a tenant entitlement determination. Organizations should validate current commercial terms, regional availability, and assigned licenses in their Microsoft 365 tenant before deployment.

  • Compromised agent detection: Identify agents exhibiting anomalous behavior that may indicate a compromised or manipulated agent (e.g., prompt injection attacks, data exfiltration patterns)
  • Malicious agent behavior detection: Detect agents performing actions inconsistent with their declared function or scope
  • Unified incident timeline: Agent activity is integrated into the unified Microsoft Defender incident timeline, enabling security analysts to correlate agent actions with user activity and other security signals
  • Cross-activity correlation: Correlate agent activity with user sign-in events, file access, and network activity to reconstruct the full sequence of an agent-involved security incident

Navigation path: Microsoft Defender portal (security.microsoft.com) > Assets > AI agents for the agent inventory and risk posture; Incidents & alerts — filter for agent-related incidents

FSI application: Agentic Copilot workflows (SharePoint agents, Copilot Studio agents) represent new vectors for insider threat and external attack. FINRA's 2026 Annual Regulatory Oversight Report specifically identifies agentic AI monitoring as a supervisory concern under FINRA Rules 3110 and 3120. Agent threat detection in Microsoft Defender provides the monitoring infrastructure to satisfy these supervisory obligations, contingent on maintaining the Microsoft Agent 365 license required as of July 1, 2026.

Security for AI — Real-Time Agent Runtime Protection

Security for AI is a Microsoft Defender capability area (Defender portal > Settings > Security for AI) that extends discovery, posture management, and runtime protection to AI agents built on Microsoft Copilot Studio, Microsoft Foundry, and Microsoft 365 Copilot, integrated through Microsoft Agent 365. Current Microsoft documentation does not evidence a prior product named "Microsoft Defender for AI"; organizations should not rely on that name for licensing or portal-navigation purposes.

Capability Description FSI Application
Real-time prompt injection detection Monitors agent interactions for direct and indirect prompt injection attempts at runtime Protects financial agents (trade documentation, client onboarding) from adversarial manipulation
Data exfiltration monitoring Detects patterns where agents are used to systematically extract sensitive data through conversational queries Identifies insider threat vectors that bypass traditional DLP controls
Agent behavioral baseline Establishes normal operational patterns for each agent and alerts on deviation Supports FINRA Rule 3110 supervisory obligations by providing continuous behavioral monitoring
Cross-platform coordination Extends runtime protection across M365 Copilot agents and Copilot Studio agents onboarded to Microsoft Agent 365 Provides unified security coverage regardless of agent authoring surface

Onboarding Security for AI:

  1. Confirm the tenant holds a Microsoft Agent 365-eligible license using the qualifying paths above and has completed Agent 365 onboarding — required for Copilot Studio and Microsoft Foundry agent coverage as of July 1, 2026
  2. Navigate to Microsoft Defender portal > Settings > Security for AI > Get started, and confirm the Security for AI Agents toggle is enabled
  3. Connect the Microsoft 365 connector (Microsoft Entra ID management events and Microsoft 365 activities) to enable investigation and advanced hunting for agent activity
  4. For Copilot Studio agents: enable the Copilot Studio real-time protection step, copy the onboarding URL from Defender, and send it to the organization's Power Platform administrator. After the administrator completes the Power Platform configuration, obtain the App ID from them, paste it into Defender, and save the connection.
  5. Under Settings > Security for AI > Policies & rules > Real-time protection, review the built-in Default audit rule and create agent-scoped custom rules for the documented detection types where blocking is required. Use Advanced Hunting telemetry to build custom detections and downstream automation for organization-specific anomalous behavior.
  6. Configure alert routing to the SOC and compliance teams

Validation tests:

  • Deploy a test agent with intentional prompt injection vulnerability; verify Security for AI detects the injection attempt
  • Simulate high-volume data extraction through a test agent; verify anomaly detection fires within the configured SLA
  • Verify that Security for AI alerts appear in the Microsoft Defender unified incident timeline alongside correlated agent activity, and that the agent appears in Assets > AI agents with a current risk level

Power Platform coordination: Copilot Studio agents require coordination between the M365 security team and Power Platform administrators to complete the App ID exchange described above. Security for AI policies defined in Defender must align with Power Platform DLP policies to avoid gaps where agent-generated data flows bypass one enforcement point but not the other.

Conditional Access App Control Integration

MDCA session controls are activated through Conditional Access App Control:

User Access Request
        │
        ▼
  Entra Conditional Access
        │
  ┌─────┴──────────┐
  │ CA Policy:      │
  │ Use Conditional │
  │ Access App      │
  │ Control         │
  └─────┬──────────┘
        │
        ▼
  ┌─────────────────────┐
  │  Defender for       │
  │  Cloud Apps         │
  │  Session Proxy      │
  │                     │
  │  ├─ Monitor mode    │
  │  ├─ Block downloads │
  │  ├─ Protect uploads │
  │  └─ Custom session  │
  │     policies        │
  └─────────┬───────────┘
        │
        ▼
  Copilot Session
  (monitored/controlled)

Documented MDCA Anomaly Detections Relevant to Copilot-Adjacent Activity

Detection Description Risk Level FSI Concern
Impossible travel Identifies user activities from geographically distant locations in an implausible time window High Potential credential theft or session hijacking affecting Microsoft 365 web access
Activity from infrequent country/region Identifies activity from locations the user hasn't recently or ever used Medium-High Potential unauthorized access to cloud sessions that can reach Copilot-grounded data
Malware detection Identifies malicious files in supported cloud storage and can support real-time upload/download governance; Microsoft documents this built-in policy as disabled by default High Malware-staged exfiltration or infected file handling in SharePoint, OneDrive, or connected SaaS apps after the policy is explicitly enabled
Unusual activities (by user) Flags unusual multiple file downloads, file shares, file deletions, impersonated activities, or administrative activities relative to the learned baseline Medium-High Bulk download or unusual share behavior around Copilot-relevant data stores
Multiple failed login attempts Identifies users with failed logins in a single session relative to the learned baseline Medium Account-compromise attempts against cloud apps in scope for Copilot governance

MDCA doesn't document dedicated per-prompt Copilot detections for query volume, topic probing, or off-hours Copilot prompting. Use Microsoft Purview audit and Insider Risk Management for prompt/response telemetry and AI-specific behavioral analysis; see Control 3.1 and Control 2.10.

Session Policy Types for Copilot

Policy Type Description Action Use Case
Monitor only Logs the routed Login activity only Audit Validate that Conditional Access App Control is redirecting the intended browser sessions
Block activities + Audit Monitors or blocks supported in-session actions such as printing, clipboard, send/share, or edit actions Audit or block Add detailed monitoring for sensitive web-session behaviors after redirection is validated
Control file download (with inspection) Monitors and can audit, block, or protect file downloads Audit / Block / Protect Govern sensitive downloads from SharePoint, OneDrive, and other supported web apps
Control file upload (with inspection) Monitors and can audit or block uploads Audit / Block Prevent unlabeled, sensitive, or malware-laden uploads in supported web apps

Copilot Surface Coverage

M365 Application Browser Session Control Anomaly Detection Activity Logging Notes
Microsoft Copilot / Copilot Chat Partial Yes Yes Browser sessions can be routed through Conditional Access App Control; use Purview audit for prompt/response telemetry
Word (web) Yes Yes Yes Browser sessions only; native clients require access controls rather than session proxy
Excel (web) Yes Yes Yes Browser sessions only
PowerPoint (web) Yes Yes Yes Browser sessions only
Outlook on the web Yes Yes Yes Browser sessions only
Teams (web) Partial Yes Yes Browser session controls apply; Teams desktop isn't supported for session controls
OneNote (web) Yes Yes Yes Browser sessions only
Loop Yes Yes Yes Browser sessions only
Copilot Pages Yes Yes Yes Web-based workflow; align with SharePoint and OneDrive download/upload governance
SharePoint and OneDrive (web) Yes Yes Yes High-value surfaces for file download/upload control where Copilot-grounded data is stored

Governance Levels

Level Requirement Rationale
Baseline Enable MDCA for the Microsoft 365 apps in scope; validate browser-session redirection with a Monitor only policy; add at least one audited session or file-control policy for the protected web sessions; review MDCA alerts weekly; review the Generative AI category in the cloud app catalog monthly; enable agent monitoring alerts in Microsoft Defender for Copilot agent deployments when the tenant has the required Microsoft Agent 365 entitlement Provides initial visibility without assuming unsupported native-client coverage or prompt-level telemetry. Shadow AI review and agent monitoring extend the control beyond first-party Copilot alone.
Recommended Configure Conditional Access App Control for the relevant Microsoft 365 browser sessions; use Block activities or Control file download/upload (with inspection) session policies with Audit or Block actions as appropriate; integrate MDCA alerts with SIEM; alert review within 4 hours for high severity; review Cloud Discovery results and unsanction high-risk generative AI apps; configure custom detections or Advanced Hunting workflows for agent anomalies in Microsoft Defender Active monitoring with supported browser-session controls and Shadow AI governance, while reserving Purview audit for Copilot prompt/response evidence.
Regulated All Recommended requirements plus: require web-only access where session controls are mandatory; use step-up authentication or blocking for elevated-risk actions; incorporate MDCA session-policy evidence, CA App Control traffic logs, and Cloud Discovery governance into examination packages; quarterly MDCA effectiveness review by security team; agent threat detection integrated into SOC playbooks with mandatory investigation SLAs (per FINRA Rules 3110/3120 supervisory requirements); Shadow AI governance policy documented with prohibited generative AI app list reviewed quarterly Full session-level governance with stronger enforcement on supported browser surfaces, explicit native-client compensating controls, and documented Shadow AI/agent governance evidence.

Setup & Configuration

Step 1: Enable Defender for Cloud Apps

Portal: Microsoft Defender portal (security.microsoft.com) > Settings > Cloud Apps > Connected apps > Conditional Access App Control apps

  1. Verify MDCA license is active for the protected users
  2. Complete initial MDCA setup if not already configured
  3. Confirm the Microsoft 365 apps you intend to govern are enabled for Conditional Access App Control

Step 2: Configure Conditional Access App Control

Portal: Microsoft Entra Admin Center > Protection > Conditional Access > Policies

  1. Create or modify a Conditional Access policy for the Microsoft 365 web apps and user groups that require session-control coverage
  2. Under Session controls, select "Use Conditional Access App Control"
  3. Use access controls separately if you must block unsupported native clients and force users into browser sessions where MDCA can apply session policies

Step 3: Create Session Policies for Copilot

Portal: Microsoft Defender Portal > Cloud Apps > Policies > Policy management > Create policy > Session policy

Policy 1: Validate Session Redirection

Setting Value
Name FSI-Copilot-Redirect-Validation
Session control type Monitor only
App filter The enabled Microsoft 365 web apps in scope (for example, Microsoft 365, SharePoint Online, or Teams web as listed in your tenant)
Action Validate redirected Login activity
Alert Optional during pilot; use primarily to confirm routing

Policy 2: Audit Sensitive File or Message Actions

Setting Value
Name FSI-Copilot-Sensitive-Session-Audit
Session control type Block activities or Control file download/upload (with inspection)
Activity source Supported browser-session activity in the Microsoft 365 web apps you selected
Content inspection Sensitivity label, SIT, malware, or trainable-classifier conditions as supported by the session control type
Action Use Audit first; add Block, Protect, or step-up authentication where the business case supports enforcement
Severity High

Step 4: Configure Anomaly Detection

Portal: Microsoft Defender Portal > Cloud Apps > Policies > Policy management

  1. Review the anomaly detection policies that are currently active in your tenant. Microsoft keeps the anomaly-detection engine on by default and, starting June 2025, transitions several legacy detections to dynamic threat detection. Note that malware detection is disabled by default and must be enabled if you want that signal.
  2. Allow for the documented seven-day learning period before treating missing anomaly alerts as evidence of a configuration problem.
  3. Prioritize detections that are most relevant to Copilot-adjacent Microsoft 365 risk, such as Impossible travel, Activity from infrequent country/region, and Malware detection
  4. Document any legacy detections that Microsoft has renamed or moved into the dynamic model so investigators do not rely on retired policy names

Step 5: Govern Generative AI Discovery and Sanctioning

Portal: Microsoft Defender portal > Cloud apps > Cloud app catalog and Cloud apps > Cloud discovery > Discovered apps

  1. Use the Cloud app catalog with the Generative AI category filter to review Microsoft risk scores for AI apps
  2. Use Cloud discovery > Discovered apps to confirm which generative AI apps are actually observed in your environment
  3. Mark prohibited apps as Unsanctioned and use Microsoft Defender for Endpoint or another supported governance stream to block them where required
  4. Mark approved apps as Sanctioned if that tag is part of your governance model and document the review cadence

Step 6: Configure SIEM Integration

# MDCA SIEM agents (generic SIEM integration) are deprecated: no new SIEM agents could be
# configured after June 19, 2025, and Microsoft has been retiring existing agents starting
# November 2025. For new integrations, use one of the following supported paths instead:
#
# 1. Microsoft Sentinel agent integration (Preview) — still supported:
#    Configure in Defender Portal > Settings > Cloud Apps > System > SIEM agents > Add SIEM agent > Sentinel,
#    or use the Microsoft Defender for Cloud Apps data connector in Microsoft Sentinel.
# 2. For other SIEM platforms, use APIs rather than a SIEM agent:
#    - Microsoft Defender XDR Streaming API (alerts and activities)
#    - Microsoft Graph Security Alerts API (List alerts_v2)
#    - Microsoft Defender XDR incidents API

Financial Sector Considerations

  • Bulk Data Extraction Detection: Financial firms face the risk of employees using Copilot to rapidly extract and aggregate sensitive data that would be difficult to assemble manually. Use MDCA to watch for unusual downloads, shares, or other cloud-app anomalies around SharePoint and OneDrive data, and use Purview Audit or Insider Risk Management for prompt-level analysis.
  • Departing Employee Monitoring: Insider risk is heightened when employees announce departures. Use HR-triggered insider-risk workflows and Purview audit review for departing-user Copilot investigations; MDCA can contribute supporting browser-session or cloud-app context where those users access Copilot-governed data through supported web sessions.
  • Trading Desk Surveillance: For proprietary trading desks, use Purview Audit, Communication Compliance, or Insider Risk Management to review prompt or response content where approved, and use MDCA as a companion signal for anomalous browser-session or file-activity context around regulated data stores. Integrate those records with the firm's trade surveillance program.
  • Shadow AI Governance via Generative AI Catalog: The MDCA catalog now contains 1,000+ generative AI apps. FSI firms face regulatory exposure when employees use unsanctioned AI tools to process customer or financial data. Use the catalog to identify generative AI tools in use, assess their risk scores, and enforce governance policies that block high-risk Shadow AI while permitting sanctioned Copilot use. GLBA §501(b) safeguard requirements extend to all tools that process customer information, including Shadow AI tools.
  • Agent Threat Detection in SOC: Copilot Studio agents and other organizational agents are subject to threat detection in Microsoft Defender, contingent on the tenant holding the Microsoft Agent 365-eligible license required as of July 1, 2026 for this coverage. FSI security teams should include agent threat alerts in SOC triage procedures. FINRA's 2026 Annual Regulatory Oversight Report specifically identifies agentic AI as a supervisory concern — SOC integration of agent threats supports FINRA Rules 3110/3120 supervisory obligations.
  • Regulatory Examination Response: During examinations, regulators may request evidence of AI monitoring capabilities. MDCA session-policy reports, CA App Control traffic logs, cloud-discovery governance actions, and agent threat detection configurations demonstrate proactive monitoring of Copilot-adjacent web sessions and Shadow AI usage. Use Purview audit logs when prompt/response-level Copilot evidence is required.
  • Teams Desktop Limitations: MDCA's Conditional Access App Control uses a reverse proxy for session-level monitoring, which works best with web applications. The Microsoft Teams desktop application is not supported for session controls, so plan compensating access controls when designing monitoring coverage.
  • False Positive Management: Financial services users may legitimately use Copilot heavily (e.g., research analysts, portfolio managers). Tune anomaly detection thresholds using a monitoring period before enforcement to establish accurate baselines per role.
  • Integration with SOC: MDCA alerts for Copilot and agent threat detection alerts should be integrated into the firm's Security Operations Center (SOC) workflow with defined response procedures for each alert type, including agent-specific investigation runbooks.
  • AI Agent Inventory and Posture Risk in Microsoft Defender: Microsoft Defender provides an AI agent inventory (Defender portal > Assets > AI agents) that calculates a risk level for each agent from its active risk indicators; agent posture-risk assessment entered public preview in July 2026. FSI organizations should integrate this inventory into their agent governance framework — it provides a native surface for identifying high-risk agents that access financial data, customer information, or regulated systems. This inventory requires a Microsoft Agent 365-eligible license as of July 1, 2026 for Copilot Studio and Microsoft Foundry agents. Configure risk classification policies to flag agents that require additional security review under the organization's third-party risk management framework.

Verification Criteria

  1. MDCA Connectivity: Verify that the relevant Microsoft 365 apps are connected or enabled and showing activity data in Defender for Cloud Apps
  2. Conditional Access App Control: Confirm at least one Conditional Access policy routes the protected Microsoft 365 browser sessions through MDCA App Control
  3. Session Policy Active: Verify that session policies are enabled and generating policy-report matches; if a policy uses Monitor only, confirm investigators understand it logs the routed Login activity only
  4. Detailed Activity Monitoring: For file downloads, uploads, clipboard, print, or send/share monitoring, confirm a Block activities or Control file download/upload (with inspection) policy exists with the correct Audit, Block, or Protect action
  5. Anomaly Detection: Confirm the tenant's dynamic anomaly detections are being reviewed after the seven-day learning period, that investigators are not relying on retired legacy-policy names, and that malware detection is enabled if the organization expects that signal
  6. Alert Routing: Verify that high-severity MDCA alerts are routed to the appropriate team (security, compliance) within the defined SLA
  7. SIEM Integration: Confirm MDCA alerts and CA App Control traffic logs are flowing to the firm's SIEM or Microsoft Sentinel; use Purview audit for prompt/response-level Copilot telemetry if required
  8. Generative AI Discovery Governance: Confirm the Generative AI category is reviewed in the cloud app catalog and that Cloud Discovery governance actions exist for high-risk generative AI apps
  9. Agent Threat Detection: Confirm the tenant holds the Microsoft Agent 365-eligible license required (as of July 1, 2026) for Copilot Studio and Microsoft Foundry agent coverage; verify agents appear with a current risk level in Defender portal > Assets > AI agents and that at least one agent-related detection workflow is configured
  10. Investigation Workflow: Verify that documented procedures exist for investigating MDCA alerts, cloud-discovery governance actions, and agent threat-detection alerts related to Copilot usage
  11. Periodic Review: Confirm that MDCA session policies, anomaly-detection handling, and generative AI governance actions are reviewed monthly (Recommended) or quarterly (Regulated) with documented results

Advisory: Security Copilot Inclusion for Eligible M365 E5 and E7 Tenants

Action Required — Phased Rollout Started November 18, 2025

Microsoft states that rollout started November 18, 2025, for existing Security Copilot customers with Microsoft 365 E5 and continues in phases for other eligible Microsoft 365 E5 and E7 tenants; it has not published a fixed completion date. Eligible tenants receive tenant-level Security Copilot inclusion through zero-click activation, with 400 Security Compute Units (SCUs) each month per 1,000 paid user licenses, capped at 10,000 SCUs each month. No Azure setup, consent flow, or manual SCU provisioning is required for that tenant-level inclusion. Affected tenants are notified through the Microsoft 365 admin center Message Center and in-product banners in Defender, Purview, Entra, Intune, and the standalone Security Copilot portal.

FSI governance implications:

  • Tenant inclusion versus agent deployment: Inclusion provisions Security Copilot for the eligible tenant and makes standalone and embedded experiences available. It does not automatically enable individual Security Copilot agents: the organization must explicitly set up and deploy each agent in the relevant Defender, Entra, Intune, or Purview experience.
  • Owner settings: Tenant owners can limit Security Copilot's access to Microsoft 365 service data and adjust Customer Data sharing preferences from Owner settings in the Security Copilot portal. This constrains data access but does not require deleting previously provisioned capacity. FSI organizations should evaluate these settings and approve agent deployments during their governance review.
  • Data processing: Security Copilot processes security event data using Microsoft's AI models; by default the tenant's Microsoft Entra geography is used for Customer Data storage, and Microsoft's human-review data-sharing settings default to Off. Organizations should verify that data processing meets internal data governance and residency requirements (see Control 2.7 — Data Residency).
  • Audit implications: Security Copilot generates its own activity records in the Purview audit log. Existing audit monitoring should be extended to capture Security Copilot activities (see Control 3.1 — Audit Logging).
  • Budget impact: Microsoft states that usage beyond the included allocation will be throttled at a future date. A pay-as-you-go option at $6 USD per SCU will begin only when Microsoft makes it available after 30 days' advance notification; do not treat throttling or this inclusion-model pay-as-you-go option as already active. Organizations should track SCU consumption and coordinate with Control 4.8 — Cost Allocation.

Recommended actions:

  1. Confirm via Microsoft 365 Message Center whether tenant-level Security Copilot inclusion has been enabled
  2. Review Owner settings and evaluate the data-sharing/data-access configuration against governance requirements
  3. Inventory, approve, and explicitly deploy only the individual Security Copilot agents required for the approved use cases
  4. Update MDCA and Defender alert routing to include Security Copilot-generated alerts
  5. Extend SIEM integration to capture Security Copilot audit events

Additional Resources