The Wire
BusinessTechnologyArtificial IntelligenceCybersecurity

Microsoft unveils first cyber AI model, agentic security platform

Microsoft unveils first cyber AI model, agentic security platform
Photo: techcrunch.com

Microsoft launched MAI-Cyber-1-Flash, a cybersecurity AI model, and Project Perception, an agentic security platform.

Why it matters: The products give Microsoft a clearer offer for enterprise security teams that need faster vulnerability finding and automated remediation. They also deepen Microsoft's commercial push into AI security at a time when rivals are racing to sell security tools built on generative AI.

  • Microsoft introduced MAI-Cyber-1-Flash, which it calls its first cybersecurity-specialized AI model, and Project Perception, a new agentic security platform.
  • Microsoft says MAI-Cyber-1-Flash is designed to find difficult vulnerabilities in complex codebases and power MDASH, its vulnerability-identification and remediation harness.
  • Project Perception uses multiple agents to automate security workflows, including identifying and fixing bugs, and it can connect with MDASH.
  • Microsoft said both products will be available in preview beginning Nov. 3, 2026.

Microsoft on July 27 launched two linked security products: MAI-Cyber-1-Flash, its first cybersecurity-specific AI model, and Project Perception, an agentic security platform built to automate parts of security work. TechCrunch reported the company introduced the tools at a small event in San Francisco.

Microsoft says MAI-Cyber-1-Flash is meant to find hard-to-spot vulnerabilities in complex codebases and feed them into MDASH, the company's multi-model vulnerability identification and remediation harness. Mustafa Suleyman, CEO of Microsoft AI, said, "We're shipping this into production immediately."

Project Perception is the broader platform layer. In Microsoft's launch post, the company said the platform uses a multi-model architecture that chooses models based on quality, reliability, latency and cost. Microsoft also said the system is integrated across Microsoft Security products so agents can move from insight to action while keeping customers in control.

MDASH is not a separate product launch. It is the internal harness Microsoft says underpins the new model and informs the platform's automation approach. Microsoft said its earlier MDASH work found 16 new vulnerabilities in the Windows networking and authentication stack, including four critical remote code execution flaws, in a May security blog post.

Microsoft said public preview for both MAI-Cyber-1-Flash and Project Perception starts Nov. 3, 2026.

By the numbers

  • 16 - vulnerabilities Microsoft said MDASH found in Windows networking and authentication
  • 4 - critical remote code execution flaws among those vulnerabilities
  • Nov. 3, 2026 - preview start date Microsoft gave for both products

Yes, but: Microsoft has not said how broadly customers will be able to use the new tools in preview or what they will cost.

What's next: Public preview for MAI-Cyber-1-Flash and Project Perception begins Nov. 3, 2026.

Based on reporting from

  • TechCrunch
  • The Verge

See how this story touches your network - open The Wire in Jane.

Open in Jane