Microsoft MDASH: AI Unveiled at Build 2026

⚡

Key Takeaways

1Microsoft has unveiled MDASH, an AI system designed to reduce the noise from security alerts by targeting exploitable vulnerabilities.

2MDASH will be integrated into tools like Defender and GitHub Code Security for a more comprehensive enterprise security control.

3MDASH achieved a score of 96.55% on the CyberGym benchmark, demonstrating its enhanced effectiveness.

💡Why it matters — The integration of MDASH could revolutionize vulnerability management, strengthening enterprise security against increasing cyber threats.

Microsoft Unveils MDASH for Enhanced Security

At the Build 2026 event, Microsoft introduced MDASH, an innovative multi-model agent-based scanning system dedicated to security, which is now moving out of the preview phase. This new tool is designed to reduce the noise of security alerts by focusing on truly exploitable vulnerabilities.

A Strategic Integration into the Microsoft Ecosystem

Microsoft announced that MDASH would be integrated into an enterprise security control plan, in connection with tools such as Defender, GitHub Code Security, Agent 365, and Purview. Aleš Holeček, Chief Security Architect at Microsoft, emphasized that AI has evolved from a mere curiosity to a production-level defense at the enterprise scale.

MDASH and Vulnerability Analysis

One of the major challenges in security automation is the signal-to-noise ratio. Automated tools often generate numerous alert signals. MDASH stands out by prioritizing exploitable risks, allowing teams to focus on real threats. It employs cost-effective models for high-volume operations.

Impressive Performance

Although Microsoft did not detail the models used by MDASH, the company claims it utilizes cutting-edge models for in-depth reasoning. MDASH recently achieved a score of 96.55% on the CyberGym benchmark, a notable improvement from its previous score of 88.45%.

Expansion of the Security Platform

Build 2026 marks the integration of MDASH into an expanded enterprise security platform. Eligible organizations can access an extended preview and integration with Microsoft Defender, as part of Microsoft's effort to secure the AI development lifecycle.

Response to Cyber Threats

Morgan Adamski from PwC US noted that cyber threats are evolving rapidly, with AI accelerating the scale and sophistication of attacks. Microsoft aims to simplify and strengthen security operations for increased organizational resilience.

GitHub Copilot and Vulnerability Remediation

Vulnerabilities in code are automatically enriched with real production signals, such as internet exposure and data sensitivity, to inform prioritization. Developers can use GitHub Copilot to apply AI-assisted fixes.

Towards a Secure Future

Microsoft positions itself as a key player in security within software development in the AI era. The company emphasizes the importance of not having to choose between innovation and security, highlighting the need for trusted systems for the future of AI. If Microsoft's tools can demonstrate exploitability and link it to remediation, this could transform enterprise vulnerability management and make organizations significantly more secure.