Microsoft Perception Is Generally Available, Bringing AI Agents to Security Teams

On November 17, 2026, Microsoft announced the general availability of the Perception platform and several AI agents for security operations. Some features, including vulnerability shielding, remain in preview.

Microsoft Perception has been generally available since November 17, 2026. Microsoft announced that its platform for coordinating specialized AI agents for security operations is ready for broader deployment. It also made several agents available for alert triage, incident investigation, detection authoring, and red-team reconnaissance.

The platform uses playbooks that, according to Microsoft, coordinate red, blue, and green agents. The company also states that significant actions should remain under human oversight. For a system working with security tools and enterprise infrastructure, this type of control is an important part of deployment.

Microsoft Perception: Which Agents Are Generally Available

Four agents intended for the blue team are generally available:

  • Alert Triage Agent for security alert triage,
  • Attack Investigation Agent for attack investigations,
  • Detection Authoring Agent for creating detections,
  • Threat Intelligence Agent for working with threat intelligence.

The Recon Agent, which Microsoft classifies as a red-team tool, is also generally available. Another generally available component is Posture Prioritization Agent.

The announcement covers security operations center activities from initial alert processing through investigation and the preparation of detection rules. Microsoft states that playbooks are intended to connect the activities of different agent types rather than have them operate as isolated tools.

What Remains in Preview for Now

Not all new features are ready for general availability. Risk Reduction Agent, Web App Pen Test Agent, integrations through MCP and API, and the vulnerability shielding feature remain in preview.

According to Microsoft, vulnerability shielding is intended to temporarily limit the exposure of vulnerable systems before a permanent fix is deployed. The company states that it requires human approval for critical steps. The feature does not replace patching: it does not remove the vulnerability, but is intended to mitigate risk while a fix has not yet been deployed.

It has not been confirmed that shielding can prevent the exploitation of all supported vulnerabilities. It is a temporary risk mitigation measure, not a replacement for a fix.

Automation Requires Access Management

Microsoft Perception expands the company’s offering for automated SOC processes and security exposure management. It may allow enterprises to automate triage, investigations, and some remediation recommendations. In real-world deployment, however, the key issues will be what permissions agents receive, which steps they can perform automatically, and how their activities can be audited.

Microsoft presents the platform as a system with human approval for significant actions. Specific settings and the scope of permissions will be particularly important for integrations with external cloud and security services.

The actual availability may depend on the license, region, supported services, and access to preview programs.

What to Watch Next

For customers, it will be relevant to see when Microsoft publishes detailed agent availability by region and licensing plan. Independent data on automated triage accuracy, false positives, and the operational benefits of individual agents will also be important.

For vulnerability shielding, questions remain about its further expansion beyond the current preview and its impact on customer operations. Microsoft makes product claims about faster detection, investigation quality, and risk reduction, but independent comparative results supporting these claims have not been provided.

Sources

Verified and updated: November 18, 2026 15:26

Sharing