Microsoft says the era of AI agents defending networks is here. At Ignite 2026 on November 17, it made Microsoft Perception, its agentic security platform, generally available.

Perception coordinates red, blue and green team AI agents. Now generally available are four blue team agents for alert triage, attack investigation, detection authoring and threat intelligence, plus a Recon Agent and a Posture Prioritization Agent. Microsoft says its Alert Triage Agent can handle about 80 percent of alerts in a typical Defender queue, based on roughly 8 million alerts.

Other pieces remain in preview, including a Web App Pen Test Agent, a Risk Reduction Agent with vulnerability shielding for flaws that lack a fix, and links to SIEM and SOAR tools through MCP and APIs. Microsoft also said Agent 365 and Entra can now detect 60 types of local AI agents on devices, such as coding assistants.

What it means: security teams can hand routine triage to agents, but they will need clear rules on what agents may change on their own and how their decisions are audited.