Daily Defender NewsThe week’s clearest administrator-impacting change is a December 2026 Unified RBAC remapping: the Microsoft Entra Security Operator role will lose default access to email threat policies and the Tenant Allow/Block List. A second URBAC update expands email-content access to alerts involving potentially malicious URL clicks. Documentation work centers on local AI-agent discovery prerequisites and the limits of what discovery proves, alongside a Defender for Cloud management-transfer procedure and Sentinel MCP triage guidance. Remaining edits are mostly metadata, prerelease notices, a removed RBAC sample table, or revised quarantine wording.
Defender XDR · Email and collaborationStarting in December 2026, Microsoft will remove default Unified RBAC access for the Microsoft Entra Security Operator role to email threat policies and the Tenant Allow/Block List. Organizations that need continued access must assign the relevant permissions manually.
Defender XDR · GeneralDefender’s expanded Unified RBAC email-content permissions include alerts related to potentially malicious URL clicks. Analysts with the appropriate permission can preview and download email from four alert types.
Endpoint · Endpoint protectionUpdated Endpoint guidance states that observing an agent process can create an inventory entry without a prompt or service permission, but discovery does not prove active use or successful access. The guidance also adds minimum Defender versions and Unified RBAC requirements for viewing inventory and the attack surface graph.
Defender for Cloud · Cloud and workloadsThe updated procedure says selecting DWU moves plan enablement and billing to the Defender portal, while the optional Management selection moves all supported cloud security capabilities. Managed in status values show where each environment is managed.
Sentinel · Cloud and workloadsThe updated Sentinel What's New guidance describes using the Sentinel MCP triage collection to discover data, list available Advanced Hunting tables, and run hunting queries in the built-in data lake.
Ahead of the December 2026 rollout, identify users who rely on inherited Security Operator access and manually assign the required permissions if they still need access. Review email-content permissions for analysts investigating malicious URL-click alerts. For local AI-agent visibility, update Defender to the documented minimum versions and verify the required Security for AI and Exposure Management permissions. When transferring Defender for Cloud management, use DWU for plan enablement and billing, add Management when all supported capabilities should move, and confirm the resulting Managed in status.
This period briefing was generated by AI from the tracked Microsoft Learn and Message Center changes.
Microsoft Defender has expanded Unified role-based access control (URBAC) permissions for email content access to include alerts related to potentially malicious URL clicks. Security analysts with the appropriate permission can now preview and download emails from four alert types, ensuring consistent access across scenarios.
The documentation now separates cloud-agent enablement from local-agent discovery. Local agents have separate onboarding and no longer require AI agent runtime protection for discovery.
The manage RBAC page no longer includes the sample table with two headers and two data rows.
Microsoft is updating the Microsoft Defender for Office 365 Unified RBAC mapping for the Microsoft Entra Security Operator role, removing default access to email threat policies and Tenant Allow/Block List. Rollout starts December 2026. Organizations must manually assign permissions if continued access is needed.
The setup guidance now separates local and cloud agent onboarding. Local agent discovery follows separate prerequisites and does not require AI agent runtime protection; runtime protection can be configured independently.
The documentation now covers local AI agent discovery on supported Windows and macOS devices, with specific preview exceptions. It updates discovery details, adds Microsoft VS Code extensions, revises MCP coverage, and expands macOS entries with preview indicators.
The documentation now distinguishes process-based discovery from VS Code extension collection on Windows and macOS. It also states that MCP server configuration discovery requires antivirus active mode and real-time protection, and removes the macOS preview note.
The documentation now states that observing an agent process can create an inventory entry without a prompt or service permission, and that discovery does not prove active use or successful access. It adds minimum Defender versions and unified RBAC permissions for viewing inventory and the attack surface graph.
The release notes now document Linux version 101.26101.0000, including an OpenSSL 3.6.4 upgrade and a fix for missing CentOS minor patch versions in the Defender portal.
The documentation now describes local AI agent discovery on onboarded Windows and macOS devices, including discovery of configured MCP servers and investigation through the AI agent inventory and advanced hunting.
The article’s author and Microsoft author metadata were updated, and its prerelease product notice was removed.
The article’s author metadata was updated, and its prerelease-product disclaimer was removed.
Updated author metadata and revised the ZAP example to state that a message moved to quarantine expires from quarantine.
The documentation now describes selecting DWU to move plan enablement and billing to the Defender portal, with an optional Management selection to move all supported cloud security capabilities. It also adds status values showing where each environment is managed.
The Microsoft Sentinel What's New page now describes using the Sentinel MCP triage collection to discover data, list available Advanced Hunting tables, and run hunting queries in the built-in data lake.
