Microsoft Opens Project Perception Agentic Cybersecurity System for Public Preview
Microsoft opened Project Perception, an agentic cybersecurity system coordinating red-team, blue-team and green-team agents, for public preview.
Why it matters: Agentic security tools are moving from analysis into controlled remediation, requiring strong governance and rollback controls.
Microsoft has opened Project Perception, its agentic cybersecurity system, for public preview.
Project Perception coordinates specialized AI agents across enterprise-security workflows. Red-team agents search for potential attack paths, blue-team agents investigate activity and evaluate risk, and green-team agents perform corrective actions.
The system connects signals from identities, endpoints, applications, data, cloud infrastructure and AI services to create a shared representation of an organization's security environment.
Its first major use case is software-vulnerability management through Microsoft's multi-model vulnerability identification and remediation workflow.
The important development is Microsoft's attempt to move cybersecurity from periodic scanning toward continuous agent-driven detection, prioritization and remediation. That could reduce time to fix vulnerabilities, but it raises governance questions around permissions, audit trails, approvals and rollback controls.