Build 2026: Microsoft's MDASH exits preview with 100+ specialized threat-hunting AI agents (www.zdnet.com)

At Build 2026, Microsoft announced the exit of its MDASH (Microsoft Security multi-model agentic scanning harness) from preview, now incorporating over 100 specialized AI agents to enhance security vulnerability management. MDASH is designed to cut through noise generated by traditional vulnerability scanners, prioritizing actionable risks and effectively triaging security alerts. By integrating MDASH with existing tools like Microsoft Defender and GitHub Code Security, Microsoft aims to streamline the security workflow for developers and security teams, helping them focus on vulnerabilities that are truly exploitable. This transition marks a significant advancement for the AI/ML community by shifting the paradigm from reactive to proactive security measures. Microsoft’s system leverages a combination of state-of-the-art reasoning models and lower-cost models, making it model-agnostic and able to adapt as needed. The latest enhancements promise to simplify security operations and foster resilience, allowing organizations to navigate the evolving threat landscape more effectively. As cyber threats continue to grow in sophistication, MDASH stands to redefine enterprise vulnerability management, ultimately enhancing trust and safety in AI software development and deployment.