Microsoft MDASH AI Tops Vulnerability Discovery Benchmarks

Microsoft's new MDASH agentic AI security system is topping vulnerability discovery benchmarks. This system, developed by Microsoft's Autonomous Code Security team, uses over 100 specialized AI agents and multiple AI models. MDASH helped researchers find 16 vulnerabilities across Windows networking and authentication components, including issues in the Windows TCP/IP stack and DNS handling. Some of these vulnerabilities were reachable over networks without requiring authentication. The system identified remote code execution flaws, denial-of-service issues, and information disclosure vulnerabilities. MDASH also scored 88.45% on the public CyberGym benchmark. The bottom line is that agentic AI is moving into high-value cybersecurity tasks, which could significantly improve defensive security.