Today, C1 is announcing support for enterprise-managed authorization (EMA), the open MCP standard for governing how AI agents access enterprise tools. Anthropic shipped the first implementation in Claude today. C1 issues the tokens, as the enterprise identity provider for it, starting now.
Enterprise-managed authorization comes to Claude. C1 is the control plane.# #
Enterprise-managed authorization lets a user authenticate once and then reach every compatible MCP server directly. One login, no per-tool authorization prompt. Underneath it is Cross-App Access (XAA), an open, OAuth-based standard for agent access. No single vendor owns it. Any provider can implement it, and C1 does. Anthropic putting it in front of millions of Claude users moves the whole ecosystem toward governed AI.
C1 is the enterprise identity provider for EMA. Users and agents authenticate to C1, and C1 issues the short-lived, scoped tokens they need to reach the MCP servers they're entitled to, with no follow-on per-tool login. Governance is enforced at the C1 control plane: fine-grained scope, session lifecycle management, and re-authentication policy, applied the moment each token is minted.
Putting MCPs and AI in front of the whole company# #
C1 makes it possible to put AI tools in front of every employee, governed from day one, at enterprise scale. Define entitlements once in C1. Every authorized user and agent gets access to the MCP servers they're cleared for, with no per-tool login and no credential sprawl.
For apps that don't support the standard, on-premises data, and legacy systems, pair it with C1's Access Gateway for fine-grained tool-call enforcement and the same governance controls, regardless of protocol. The same governance layer covers both. No architecture trade-offs. Every system covered.
## Driving the AI transformation, securely[#](#driving-the-ai-transformation-securely)
Connecting agents to enterprise systems shouldn't mean choosing between speed and security. C1 enforces the controls teams need on every MCP connection: fine-grained scope, runtime tool enforcement, session lifecycle management, and re-authentication policy, all before an agent touches a system. For employees, that governance is frictionless. Authenticate once to C1 and get an SSO-like experience across every compatible MCP server: one authentication flow, direct access, no per-tool login, no credential sprawl. With C1, the fastest path to AI adoption is the safest path.
Why C1?# #
C1 enforces governed access for every agent and every human from one place: through the standard wherever the agent's harness supports it, and through the gateway everywhere else. Revoke an agent's access and C1 stops issuing new tokens immediately. Remove access to an MCP server and enforcement is instantaneous. Every policy check, every access decision, and every token issuance lands in one audit trail. The answer to "what did that agent reach, and who signed off?" is a governance record, not a forensic reconstruction from logs. Controls and visibility for your AI agents, enforced through a single control plane.
To the security and identity teams building ahead of the curve: this is what governed AI deployment looks like. The agents are already reaching data, and teams need a single, unified way to manage that access. Reach out to book a demo.