Microsoft responds to security challenges facing code, AI agents, and models

Microsoft announced a set of AI-driven security tools and previews focused on vulnerability discovery and securing AI agents and models, including MDASH (a multi-agent vulnerability discovery system integrated with Microsoft Defender), Defender–GitHub Code Security integration for context-rich code findings, Agent 365 and MXC SDKs for agent isolation and observability, Purview data protection and runtime DLP for agent prompts, and Defender AI model scanning to inspect models before deployment.