Microsoft Unveils Project Perception AI Security
Microsoft has introduced Project Perception, an agentic security system it describes as designed for the realities of AI, turning signals into real-time protections and using AI to defend against AI. Microsoft argues that the approaches built for a world of human actors cannot keep pace with AI, agents and machine-speed attacks, and that security needs a new Cyber Stack that can continuously perceive risk, reason across context and act at machine speed while keeping humans in control. Project Perception coordinates three classes of specialized agents in a closed loop: red team agents identify potential paths to compromise, blue team agents investigate and determine what represents meaningful risk, and green team agents take corrective actions and strengthen defenses. It uses a multi-model architecture that combines frontier and specialized cyber models. The first scenario is software vulnerability management, with MAI-Cyber-1-Flash running inside MDASH, which Microsoft says delivers 96 percent on the CyberGym benchmark, 12 points above Mythos, with almost 50 percent cost savings versus the current MDASH configuration. Project Perception enters public preview on August 3.
Transcript
Microsoft just announced Project Perception, an agentic security system built to defend against machine speed attacks.
Microsoft says attackers generate exploits faster and scale campaigns further. Perception turns signals into real time protections, using AI to defend against AI.
Three agent teams work in a closed loop. Red agents find paths to compromise. Blue agents judge meaningful risk. Green agents take corrective actions and strengthen defenses.
On the CyberGym benchmark, MAI Cyber One Flash inside MDASH scores ninety six percent, twelve points above Mythos, with almost fifty percent cost savings. Public preview starts August third.
Inboxsmith helps small businesses handle calls and messages so nothing gets missed. Please like and subscribe for more news.
Sources
Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:
- Microsoft introduced Project Perception on July 27, 2026, describing it as a new agentic security system designed for the realities of AI that turns signals into real-time protections, using AI to defend against AI.(Microsoft official announcement)
- Microsoft says attackers can generate exploits faster, scale campaigns further, and operate with unprecedented efficiency, and that approaches built for a world of human actors cannot keep pace with machine-speed attacks.(Microsoft official announcement)
- Project Perception brings together signals, context, models and specialized agents into a continuously learning system of defense, keeping humans firmly in control.(Microsoft official announcement)
- Project Perception coordinates three classes of specialized agents that form a closed-loop system: red team agents identify potential paths to compromise before an attacker can exploit them, blue team agents investigate and determine what represents meaningful risk, and green team agents take corrective actions and strengthen defenses.(Microsoft official announcement)
- The first scenario is software vulnerability management, using MAI-Cyber-1-Flash inside MDASH, Microsoft's software-vulnerability multi-model team of agents.(Microsoft official announcement)
- Microsoft says MDASH with MAI-Cyber-1-Flash delivers 96% on CyberGym, an industry leading benchmark, 12 points above Mythos.(Microsoft official announcement)
- Microsoft says the same configuration delivers almost 50% cost savings compared with the current MDASH configuration in market today.(Microsoft official announcement)
- Project Perception enters public preview on August 3, 2026.(Microsoft official announcement)