Microsoft Unveils AI-Powered Cyber Stack to Combat Evolving AI Threats
Microsoft introduces a new three-layer cyber stack designed to leverage AI for defense against increasingly sophisticated AI-driven attacks, featuring Microsoft Perception for orchestrated agentic security.

The cybersecurity landscape is rapidly transforming as frontier AI models become adept at discovering and exploiting software vulnerabilities, a capability that is quickly becoming accessible to a wider range of actors. This acceleration, exemplified by the swift weaponization of critical PHP vulnerabilities, signifies a shift where attackers can exploit flaws faster than defenders can patch them. Microsoft highlights that AI is moving beyond assisting attackers to autonomously planning and executing entire campaigns, posing a significant challenge to traditional security measures.
In response to these evolving threats, Microsoft is introducing a new cyber stack engineered for the "agentic era." This stack is built on the principle of fighting AI with AI, aiming to provide defenses that can operate at machine speed. The new architecture comprises three distinct layers: Signals, Context, and Agents. The Signals layer acts as the foundation, collecting over 165 trillion daily signals from sensors and actuators embedded across Microsoft's product suite, including Defender, Entra, Intune, Purview, and Sentinel, as well as the broader ecosystem through over 500 connectors.
The Context layer processes these vast amounts of signals, integrating them with customer-specific environment data. This layer transforms raw data into an "agent-ready, real-time understanding" of the digital estate, optimized for precision, recall, and efficiency. This comprehensive understanding is crucial for enabling security agents to operate effectively and make informed decisions in complex environments.
The top layer consists of specialized security agents, which collaborate through playbooks orchestrated by a harness. This multi-model approach ensures that the most appropriate AI model is used for each security task, enabling efficient and scalable threat response. These agents are designed to work like a coordinated team, addressing security outcomes rather than just individual tasks.
At Microsoft Ignite 2026, the company is launching Microsoft Perception, now generally available, which orchestrates red, blue, and green team agents. Red team agents simulate attacks, blue team agents detect and investigate threats, and green team agents focus on remediation. This closed-loop system aims to help security teams operate at machine speed, automate processes, and maintain control by setting guardrails and approving high-impact actions.
Microsoft is also investing in foundational security operations capabilities. The Integrated Security Operations Center (ISOC) within Microsoft Defender, now generally available for core functions, unifies SIEM, UEBA, threat intelligence, and threat protection. This integration aims to streamline security operations and facilitate the transition to an agentic Security Operations Center (SOC).
New cost management dashboards and estimators are also being introduced to help organizations track AI usage, forecast spending, and link costs to tangible security outcomes. This focus on cost transparency is vital as organizations increasingly adopt AI-driven security solutions.
The introduction of this new cyber stack signifies Microsoft's strategic pivot to address the unique challenges posed by AI-powered threats. By building defenses that mirror the speed and sophistication of AI-driven attacks, Microsoft aims to equip organizations with the tools necessary to secure their digital environments in the emerging agentic era.