Microsoft AI Finds 16 Windows Security Flaws

Microsoft is using agentic AI to find security flaws in Windows before attackers can exploit them. The company's MDASH system helped researchers identify 16 vulnerabilities. This includes four Critical flaws that could allow remote code execution. What's interesting is that MDASH uses over 100 specialized AI agents. These agents examine code from different angles, identify potential bugs, and then test if those findings are real. This multi-model approach helps ensure that cybersecurity teams get actionable insights, not just a list of possible problems. MDASH was built by Microsoft’s Autonomous Code Security team and is now used by Microsoft security engineering teams. It's also being tested by a small group of customers. The identified flaws were included in Microsoft’s May 12, 2026 Patch Tuesday update. This shows how AI is being used to inspect and challenge code, finding weaknesses before they can be exploited. This development highlights the evolving role of AI in cybersecurity, software development, and technical skills training.