Cybercriminals are increasingly turning to AI to launch faster and more sophisticated attacks, and Microsoft believes the security industry needs to evolve just as quickly. With that in mind, the company has announced Project Perception, a new AI powered cyber security platform that uses multiple AI agents to identify threats, investigate them and automatically strengthen an organisation's defences.
Unlike traditional security tools that mainly alert security teams, Project Perception is designed to analyse risks, understand the bigger picture and take action where needed. Microsoft says the platform combines security signals, contextual information, AI models and specialised AI agents into a single cyber stack capable of defending organisations against AI powered attacks.
At the heart of the platform are three categories of AI agents, each handling a different part of the security process. The Red team agents behave like attackers by searching for weaknesses that hackers could exploit. The Blue team agents then investigate those findings, analyse their context and identify the most serious risks. Finally, the Green team agents step in to fix vulnerabilities and improve the organisation's security before attackers can take advantage of them.
Microsoft says these AI agents work together as a "closed loop system" that continuously discovers weaknesses, evaluates risks and strengthens cyber defences, creating an ongoing cycle of protection rather than reacting only after an attack has happened.
However, Project Perception is more than just a collection of AI agents. Microsoft has also introduced a redesigned cyber stack to support them. The foundation of the platform consists of signals and sensors that monitor endpoints, user identities, cloud environments, applications and data. This information is then passed to a context layer that continuously enriches security intelligence, giving AI agents the background they need to understand what is happening across an organisation's network.
Image Credits: Microsoft
Above this sits a layer of specialised AI models that reason over potential threats using different areas of expertise. The final layer includes orchestrators, specialised AI agents and actuators that decide how those insights should translate into real world defensive actions, allowing the platform to respond much more quickly than conventional security systems.
Microsoft has confirmed that Project Perception will be available in public preview from August 3.
Explaining why the company believes a new approach to cyber security is necessary, Microsoft said AI has dramatically changed the way attackers operate. "Attackers can generate exploits faster, scale campaigns further, and operate with unprecedented efficiency. The approaches built for a world of human actors cannot keep pace with a world of AI, agents, and machine-speed attacks," the company said.
ALSO READ: Amazon Great Freedom Sale 2026 Announced: Sale Dates, HDFC Discount And Offers Revealed
As AI continues to reshape the cyber threat landscape, Microsoft believes platforms like Project Perception could help organisations keep pace by detecting threats earlier, responding faster and continuously improving their security without relying solely on manual intervention.