Microsoft launches Project Perception agentic cybersecurity platform

Microsoft has introduced Project Perception, an agentic cybersecurity system designed to detect, assess and respond to threats at machine speed.

The platform combines security signals, organizational context, AI models and specialized agents into a continuously learning defense system. Microsoft said Project Perception will enter public preview worldwide on August 3.

Project Perception coordinates three categories of agents. Red team agents simulate attacks and identify potential routes into an organization. Blue team agents investigate activity and determine which risks require attention. Green team agents take corrective action and strengthen defenses across the environment.

The system uses a multi model architecture that selects different frontier and specialized cybersecurity models based on quality, reliability, speed and cost. Microsoft said this approach is intended to provide continuous protection without relying on a single AI model.

Its first application will focus on software vulnerability management through MDASH, Microsoft’s team of vulnerability scanning agents. 

The system will incorporate Microsoft’s MAI Cyber 1 Flash model, which scored 96% on the CyberGym benchmark, 12 points above Mythos. Microsoft said the configuration also reduced costs by nearly 50% compared with the existing MDASH system.

Project Perception will integrate with Microsoft’s existing security products across identities, endpoints, applications, data, cloud infrastructure and AI systems. The company said the agents will be able to turn security insights into protective actions while leaving human operators in control.