The new Project Perception turns signals into real-time protections and uses AI to defend against AI, the company said in a Monday (July 27) blog post.
“Project Perception is based on a simple idea: effective defense requires continuous understanding of how an attacker sees the world, how a defender evaluates risk and how protections are improved over time,” Microsoft said in the post.
Perception coordinates three classes of specialized agents, including red team agents that identify potential vulnerabilities before an attacker can exploit them, blue team agents that investigate and determine what represents meaningful risk, and green team agents that take corrective actions and strengthen defenses, according to the post.
“Working together, these agents form a closed-loop system that continuously discovers, evaluates and improves an organization’s security posture,” the post said.
Project Perception uses a multi-model architecture that encompasses frontier and specialized cyber models and optimizes for quality and cost, per the post.
“As part of this multi-model strategy, we are committed to bringing customers the best models for each security task, including innovating with our own specialized models,” Microsoft said in the post.
Together with agents and models, Project Perception includes signals and sensors, security context, a harness that coordinates the agents and models, and actuators that turn decisions into protection, according to the post.
“Together, these layers create a continuous learning system that can understand risk, adapt to changing conditions and improve security outcomes over time,” the post said.
It was reported July 15 that Microsoft’s cybersecurity business was developing more AI security products, cutting back on some of its more traditional security products and consolidating engineering teams.
The report said Microsoft was making these changes to better respond to customer demand for solutions to the threat of AI-powered hacks, and to capture some of the spending that is going to AI firms Anthropic and OpenAI.
The PYMNTS Intelligence report “Where Payments Decisions Happen: How Issuer Data Is Powering the Next Era of Commerce” found that 42% of issuers said AI has helped them save more than $5 million from fraud attempts in recent years.
