TITLE: Anthropic Brings Enterprise IT Controls to Claude's Tool Connections DATE: 2026-06-22 COMPANY: Anthropic TOPIC: Agent Systems SUMMARY: Anthropic launched Enterprise-Managed Authorisation for Claude's MCP connectors on 18 June 2026, allowing IT administrators to provision tool access organisation-wide through Okta, the enterprise identity platform. Employees now inherit connector access automatically on their first login rather than having to authenticate each tool individually, with supported integrations including Asana, Atlassian, Figma, Canva, and Granola across Claude chat, Claude Code, and Claude Cowork. WHAT CHANGED: Anthropic launched Enterprise-Managed Authorisation (EMA) for Model Context Protocol (MCP) connectors in Claude on 18 June 2026. The update marks a meaningful shift in how businesses can manage Claude's access to external tools, moving from a model where each employee configured their own connections to one where IT controls access centrally. Under the previous model, each employee who wanted to use Claude alongside tools like Asana, Atlassian, or Figma had to individually authenticate those connections through their own Claude account. For teams with dozens or hundreds of employees, this created a consistent adoption problem: low setup rates, inconsistent permission scopes across individuals, and no central visibility into which tools Claude was accessing or on whose behalf. Enterprise-Managed Authorisation addresses this through integration with Okta, the enterprise identity platform. Administrators configure tool access once within Okta, scoped to the relevant team or role groups they manage. When an employee logs into Claude, their connector access is inherited automatically with no manual authentication required. When that employee leaves the organisation and is deprovisioned in Okta, their connector access revokes quickly rather than lingering on stale tokens. At launch, the feature supports seven connectors: Asana, Atlassian, Canva, Figma, Granola, Linear, and Supabase. Slack support is listed as coming soon. The feature is available in beta for Team and Enterprise plan subscribers and works across Claude chat, Claude Code, and Claude Cowork. Anthropic also launched companion connector observability tooling, giving administrators a dashboard view of adoption, errors, latency, and usage across all active connectors. WHY IT MATTERS: The MCP standard has grown rapidly since its introduction in late 2024, reaching 97 million installs by early 2026. The number of tools Claude can connect to is large and still growing, which means central IT management of those connections is now a practical necessity rather than a convenience. For businesses deploying Claude across a team of 20 or more people, individual setup requirements have been the single biggest adoption barrier in practice. Removing this friction removes the primary reason most organisations have kept Claude siloed to a small group of power users. The Okta integration means AI tool access is now governed by the same identity and access management system most mid-size businesses already use to control access to Salesforce, Atlassian, and Google Workspace. Claude becomes part of the governed IT stack rather than a separate self-service tool. Connector observability gives IT administrators the usage data they need to justify or adjust AI tool investment. Shadow AI use remains high in most organisations; a central visibility layer helps identify where employees are working around approved tools. Fast deprovisioning through the identity provider reduces the security exposure of former employees retaining AI tool access, a risk that has grown as companies scale up AI use across more systems. Additional identity providers beyond Okta are expected in subsequent releases, which will extend this capability to businesses using Microsoft Entra, Google Workspace Identity, or other platforms. DAVID & GOLIATH ANALYSIS: The adoption gap in enterprise AI is not a capability problem. It is a friction problem. Businesses that have invested in Claude subscriptions often see a handful of committed users extracting real value while the majority of the team never gets past the setup phase. That gap does not exist because Claude is hard to use once you are in. It exists because the path from "we have a licence" to "every relevant employee has Claude connected to the tools they actually use every day" has been longer than most IT teams can absorb alongside their other priorities. Enterprise-Managed Authorisation changes that equation. For a business using Okta, the deployment work now happens once at the administrator level. The team wakes up with their tools already connected. They do not need to know what MCP is. They do not need to separately authenticate Asana or Figma. They open Claude and their working context is already there. For businesses of 20 to 200 people, this is a practical inflection point. The question is no longer "can we get Claude to work with our tools" but "what should our team actually do with Claude now that it can access everything they work in." That is a considerably more interesting problem to be solving. Organisations that move through this setup barrier first will have a genuine head start over competitors who are still managing individual authentication tickets. RELEVANT SYSTEMS: Secure AI Brain, Employee Amplification Systems SOURCE URL: https://davidandgoliath.ai/daily-ai-briefing/anthropic-claude-enterprise-managed-mcp-authorization-okta FEED URL: https://davidandgoliath.ai/daily-ai-briefing/feed --- Published by David & Goliath | https://davidandgoliath.ai Daily AI Briefing: one AI development per day, decoded for business operators. This is a structured companion file optimised for LLM retrieval and citation.