Microsoft · Agentic security

Project Perception

Governed intelligence profile for Project Perception

First observed 2026-08-27 · Last verified 2026-08-27 ·FRESH · ASI-1.0 · ASI-TAXONOMY-1.1

Lifecycle

UNKNOWN — no evidence-backed lifecycle event recorded.

Agents

UNKNOWN — no named agent is established by current evidence.

Autonomy and control

UNKNOWN — autonomy facets have not been established.

Project Perception documents human sign-off for every high-impact action, with defender-set objectives and guardrails.

AFFIRMED

Architecture

Project Perception documents a red/blue/green multi-agent architecture coordinated by orchestrated workflows and an orchestration harness.

AFFIRMED

Capabilities

UNKNOWN — no capability assertion is established.

Evidence conflicts

No open conflict is recorded for this system.

Recent changes

APPROVAL_CHANGE

Project Perception documents human sign-off for every high-impact action, with defender-set objectives and guardrails.

AUTHORIZATIONOPERATIONSRISK

ARCHITECTURE_CHANGE

Project Perception documents a red/blue/green multi-agent architecture coordinated by orchestrated workflows and an orchestration harness.

ARCHITECTUREEVALUATIONRISK

AUTONOMY_FACET_CHANGE

Project Perception assigns work execution to agents while reserving judgment to humans.

AUTHORIZATIONEVALUATIONRISK

AVAILABILITY_LIFECYCLE_CHANGE

Project Perception is in preview, initially delivered through Microsoft Defender, with broader Microsoft Security expansion planned.

DEPLOYMENTEVALUATIONPROCUREMENT

AVAILABILITY_LIFECYCLE_CHANGE

Microsoft Learn identifies Project Perception as Limited Public Preview.

DEPLOYMENTPROCUREMENT

ARCHITECTURE_CHANGE

Project Perception combines signals, context, models, and red/blue/green agents through coordinated playbooks.

ARCHITECTUREEVALUATIONRISK

AUTONOMY_FACET_CHANGE

Project Perception supports autonomous execution initiated through reusable playbooks.

AUTHORIZATIONOPERATIONSRISK

CONTROL_CHANGE

Operators can approve or reject requested agent actions, stop sessions, and provide alternative guidance.

AUTHORIZATIONOPERATIONSRISK

Unknowns

Any evidence-panel row marked UNKNOWN is not a negative capability claim. Any NONE_FOUND row means the current governed corpus contains no qualifying evidence.

Sources

NONE_FOUND

Compare claims with evidence categories