Microsoft and GitHub are making a significant commitment to the future of AI interoperability by joining the steering committee for Anthropic's Model Context Protocol (MCP), announced during Microsoft's Build 2025 conference on May 19.
MCP, often described as a "USB-C port for AI applications," provides a standardized way for AI models to connect with external data sources and tools. The lightweight, open protocol enables two-way connections between AI-powered applications and various data repositories, business tools, and development environments.
Microsoft's adoption represents a major endorsement for MCP, which has rapidly gained industry acceptance since its introduction in late 2024. Both OpenAI and Google announced support for the protocol earlier this year, with OpenAI CEO Sam Altman noting that "People love MCP" and Google DeepMind CEO Demis Hassabis describing it as "a good protocol rapidly becoming an open standard for the AI agentic era."
As part of their commitment, Microsoft and GitHub have already made significant contributions to the MCP ecosystem. Microsoft's identity and security teams collaborated with Anthropic to design an updated authorization specification, allowing MCP-connected applications to more securely access services like cloud storage and subscription plans through trusted sign-in methods. Meanwhile, GitHub worked with the MCP steering committee to develop a registry service that enables developers to implement centralized repositories for MCP server entries.
Microsoft is also integrating MCP deeply into its product ecosystem, with support announced across GitHub, Copilot Studio, Dynamics 365, Azure, and Windows 11. The Windows integration is particularly significant, as it will expose core system functionalities like file system access to AI models, enabling more intuitive and automated functions within applications.
Security remains a top priority as Microsoft implements MCP. The company is developing a comprehensive security architecture for Windows 11 that includes a proxy to mediate all MCP interactions, baseline security requirements for MCP servers, and runtime isolation through granular permissions. These measures aim to address potential vulnerabilities while maintaining an open ecosystem for developers.