Microsoft is enhancing Windows 11’s security with the Model Context Protocol (MCP), a new standard for safe communication between AI agents and applications. Announced at Microsoft Build 2025, MCP aims to facilitate secure, interoperable computing while addressing emerging threats. With robust security measures, Windows 11 prioritizes user control and safety in the evolving landscape of AI technology.

Securing the Model Context Protocol: A Safer Future for Windows Users
At Microsoft Build 2025, a significant announcement was made regarding the Model Context Protocol (MCP). This protocol aims to enhance security in AI-driven applications on Windows 1
What’s New with MCP?
The Model Context Protocol is a lightweight, open standard that facilitates secure communication between AI agents and applications. Essentially, it operates as JSON-RPC over HTTP. By allowing seamless orchestration across local and remote services, MCP enables developers to build once and integrate everywhere.
“MCP opens up powerful new possibilities — but also introduces new risks.”
Windows 11 will support developers in creating intelligent applications that leverage MCP and generative AI capabilities. An early preview of these capabilities will be available for developers soon.
Major Updates in Security
As MCP expands, security becomes paramount. Without robust controls, vulnerabilities could arise. For instance, an improperly configured MCP server might expose sensitive functionalities or allow unauthorized access.
Key security risks include:
- Cross-Prompt Injection (XPIA): Malicious content can manipulate agent instructions.
- Authentication Gaps: Current standards are inconsistent, making security unpredictable.
- Credential Leakage: Agents with full user privileges risk exposing sensitive tokens.
“Security is not a one-time feature — it’s a continuous commitment.”
Windows 11 aims to provide a secure environment while evolving with emerging threats. The MCP security architecture emphasizes user control and transparency in operations performed on their behalf.
What’s Important to Know
Windows 11 will implement several security controls for MCP:
- Proxy-Mediated Communication: All interactions will route through a trusted Windows proxy for centralized policy enforcement.
- Tool-Level Authorization: Users must approve each client-tool pair, ensuring they remain in control.
- Central Server Registry: Only MCP servers meeting baseline security criteria will be listed, enhancing trust.
These measures aim to prevent attacks like tool poisoning while fostering a diverse ecosystem of MCP servers.
Looking Ahead
Microsoft is committed to evolving its security measures as MCP capabilities expand. Collaborations with industry partners will ensure that security keeps pace with innovation. The future of AI on Windows is not just powerful — it’s safe.
From the Windows Blog