Microsoft Launches Project Perception to Detect Vulnerabilities Before Exploitation

Microsoft has launched Project Perception, a new AI-powered cybersecurity platform designed to identify vulnerabilities before attackers can exploit them. The platform continuously detects risks, reasons across complex environments, and helps organizations take proactive action to strengthen their security posture.

Microsoft mentioned that traditional security tools and human-led processes cannot keep pace with the speed, scale, and complexity of AI-powered cyber threats. As attackers increasingly use AI to automate reconnaissance, discover vulnerabilities, and launch sophisticated attacks, organizations need a defense system that can continuously monitor risks, analyze vast amounts of security data, and respond in real time.

“The physics of cybersecurity are changing,” explained Hayete Gallot, Executive Vice President of Microsoft Security. “Autonomous systems can now reason, adapt and operate continuously. At the same time, the cost of offense is falling, while the volume, velocity and complexity of what must be secured continues to grow.”

Project Perception aims to solve this problem by combining AI agents, contextual intelligence, and automated actions into a continuously learning security platform that helps defenders identify threats earlier, prioritize the most critical risks, and strengthen protections before attackers can exploit weaknesses.

The role of Red, Blue, and Green team agents

Microsoft highlighted that Project Perception uses three types of specialized agents that work together as a coordinated defense system. Red team agents proactively search for vulnerabilities and identify potential attack paths that could be exploited by malicious agents. Moreover, blue team agents analyze findings, investigate threats, and assess which risks require attention. Green team agents then apply corrective actions to reduce those risks and strengthen security controls across the environment.

Microsoft’s Project Perception relies on a real-time understanding of an organization’s users, devices, applications, relationships, and risks. This context helps AI agents make more accurate and efficient decisions. Microsoft argues that no single AI model is ideal for every security task. Project Perception chooses different models based on factors such as accuracy, reliability, response speed, and cost.

Microsoft emphasized that Project Perception follows its Responsible AI principles and includes existing enterprise-grade security, compliance, governance, and operational safeguards. This platform will be available in public preview on August 3.

MAI-Cyber-1-Flash boosts MDASH performance

Last but not least, Microsoft is enhancing vulnerability management by integrating its specialized MAI-Cyber-1-Flash AI model into MDASH. This is a multi-agent system designed to identify software security weaknesses. MDASH uses multiple agents that work together to analyze software, discover vulnerabilities, and help security teams address potential risks more efficiently.

Microsoft notes that the new configuration achieved a 96% score on the CyberGym benchmark, which outperformed the Mythos model by 12 percentage points. It also reduced operating costs by around 50% compared to the current MDASH setup.