en.Wedoany.com Reported - At the Build 2026 conference, Microsoft announced that its MDASH (Microsoft Security Multi-Model Agent Scanning Framework) security capabilities have been integrated into a complete enterprise security control plane, connecting products such as Defender, GitHub Code Security, Agent 365, and Purview. This framework filters security alerts from continuous noise, converting them into high-confidence warnings that directly point to exploitable vulnerabilities.

MDASH is essentially an agent AI system used for vulnerability triage. Instead of overwhelming security teams with noise from extensive scanning, it prioritizes actionable risks, helping teams focus on exploitable vulnerabilities. The system orchestrates a pipeline of over 100 specialized AI agents, using model ensembles to discover, verify, and demonstrate the exploitability of codebases written in popular programming languages.

According to Microsoft's Chief Security Architect, Aleš Holeček, AI vulnerability discovery has transitioned from a research curiosity to enterprise-scale production-grade defense, with the lasting advantage lying in the agent system surrounding the models, rather than any single model itself. Microsoft stated that this design allows for trade-offs between speed, recall, and cost, while minimizing dependence on any given model.

In terms of technical performance, Microsoft reported that MDASH recently achieved a score of 96.55% on the CyberGym benchmark, up from 88.45% in the initial announcement last month. Meanwhile, MDASH is now available in an extended preview to eligible organizations and already includes Microsoft Defender integration.

Additionally, Microsoft Defender and GitHub Code Security are being integrated to bring runtime context into developer and security workflows. Vulnerabilities discovered in code are automatically enriched with real production signals, such as internet exposure and data sensitivity, to inform prioritization. Developers can then address issues using AI-assisted fixes generated, assigned, and verified through GitHub Copilot autofix and GitHub Copilot cloud agents.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com