Microsoft unveils MDASH, its AI agent-driven security platform — and it’s already spotted a host of new Windows flaws

In a significant advancement in cybersecurity, Microsoft has introduced MDASH, an innovative platform designed to enhance vulnerability discovery through the power of artificial intelligence. This cutting-edge tool, developed by the company’s Autonomous Code Security Team in collaboration with the Windows Attack Research and Protection group, orchestrates a network of over 100 specialized AI agents to meticulously scan for potential security flaws.

Key Discoveries and Features

MDASH has already made a notable impact, identifying 16 previously unknown vulnerabilities across various Windows components. Among these findings are four critical remote code execution (RCE) vulnerabilities, specifically CVE‑2026‑33827 in tcpip.sys and CVE‑2026‑33824 in IKEv2, which were addressed in the May updates. This proactive approach not only underscores the platform’s effectiveness but also highlights Microsoft’s commitment to enhancing the security of its operating systems.

What sets MDASH apart is its impressive accuracy; during testing, it achieved zero false positives, a feat that speaks volumes about its reliability. Currently, the platform is being utilized internally within Microsoft and is also available in a private preview, allowing select partners to experience its capabilities firsthand.

As the landscape of cybersecurity continues to evolve, tools like MDASH represent a vital step forward in safeguarding digital environments against emerging threats.

Winsage
Microsoft unveils MDASH, its AI agent-driven security platform — and it's already spotted a host of new Windows flaws