AI Model Context Protocol Adds Centralised Auth for Enterprise

2026-07-06 · Source: InfoQ · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Cybersecurity & Data Privacy · Depth: Intermediate, short

Summary

The Model Context Protocol (MCP) team announced on July 06, 2026, that its Enterprise-Managed Authorisation (EMA) extension has reached stable status. This extension introduces a centralized method for organizations to manage access to MCP servers through their existing identity providers, aiming to replace individual per-server consent prompts with a streamlined, zero-touch user experience. Users can now sign in once and access approved MCP servers without further setup. Adopted by Anthropic, Microsoft, and Okta, EMA addresses a major pain point in enterprise MCP deployments by moving authorization decisions to the identity provider. The architecture utilizes an Identity Assertion JWT Authorisation Grant (ID-JAG) to issue access tokens, separating identity policy from the tool call itself. While EMA centralizes connection-level control, it explicitly does not provide runtime authorization for individual actions, requiring separate organizational controls for in-system activities. Support is growing across the ecosystem, including Anthropic's Claude layers and Visual Studio Code, alongside server-side implementations by Asana, Atlassian, and others.

Key takeaway

For AI Architects or MLOps Engineers deploying AI agents in enterprise environments, the Model Context Protocol's Enterprise-Managed Authorisation extension simplifies access management significantly. You should evaluate integrating EMA with your existing identity provider, such as Okta, to enable single sign-on for MCP servers, reducing user friction and centralizing policy enforcement. Be aware that while EMA controls server access, you still need separate mechanisms for runtime authorization of specific actions within systems.

Key insights

Centralized authorization for AI model servers via enterprise identity providers streamlines access and enhances security.

Principles

• Centralized identity providers can manage access to AI model servers.
• Separate identity policy from tool calls for enhanced security.
• Connection-level authorization differs from per-action runtime control.

Method

An Identity Assertion JWT Authorisation Grant (ID-JAG) is exchanged for an access token by the MCP server's authorization server.

In practice

• Integrate EMA with existing enterprise identity providers.
• Implement single sign-on for multiple MCP servers.
• Establish separate controls for in-system agent actions.

Topics

• Model Context Protocol
• Enterprise Authorization
• Identity Providers
• AI Agent Security
• Access Management
• OAuth

Best for: CTO, VP of Engineering/Data, Director of AI/ML, MLOps Engineer, AI Architect, AI Security Engineer

Related on AIssential

• Centrally manage authorization for MCP connectors — Centralized IdP-driven authorization for AI connectors streamlines access, enhances security, and simplifies enterprise deployment.
• Building an AI-Powered Contract Price Adjustment Agent with Docusign MCP and Microsoft Foundry — The Model Context Protocol (MCP) enables AI agents to seamlessly integrate with and orchestrate enterprise services.
• Unity AI Gateway: How to Connect Agents to External MCPs Securely — Unity AI Gateway simplifies secure agent access to external tools by centralizing authentication and governance.
• Turning AI Assistants Into Strategy Execution Partners For Enterprise Businesses — The Model Context Protocol (MCP) enables AI assistants to directly access live business data, automating strategy execution analysis and reporting.
• What is Model Context Protocol? A practical guide to MCP - Cohere — Model Context Protocol (MCP) standardizes AI application integration with enterprise systems for data access and action execution.
• Embed Amazon Quick Suite chat agents in enterprise applications — Securely embedding conversational AI into enterprise applications can be streamlined with a comprehensive AWS-based solution.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by InfoQ.