Share with your CISO
C1 is betting that the messiest security problem in agentic AI, credential sprawl across every tool an AI agent touches, gets solved at the identity layer, not the application layer. The company now supports enterprise-managed authorization (EMA), an open standard Anthropic shipped inside Claude on June 18, 2026. C1 issues short-lived, scoped tokens so Claude agents authenticate once and reach only the tools they’re entitled to. For everything outside the standard, C1’s Access Gateway enforces the same policy. One control plane, one audit trail.
What this means for your business
The credential-sprawl problem is real and it compounds fast. Every AI agent that connects to a new enterprise tool, whether Salesforce, a ticketing system, or an internal data store, historically requires its own credentials, its own session, and its own audit log. At ten agents that’s annoying. At a thousand it’s ungovernable, and the blast radius of a single compromised agent session becomes an organization-wide exposure. CISOs who’ve already deployed any agentic tooling are probably on the wrong side of this problem right now, whether they know it or not.
The architecture C1 is describing is essentially OAuth for agents, where tokens are short-lived (expiring in minutes, not days), scoped to specific tools, and revocable from a single control plane. That’s a meaningful improvement over the current default, which is long-lived API keys stored somewhere on a developer’s laptop or hardcoded into a pipeline. The open Cross-App Access standard underneath EMA matters because it means the governance model isn’t locked to C1: if the standard gets broad adoption, enterprises don’t inherit a new single-vendor dependency to escape the old one. That caveat is worth watching.
C1 is a vendor with an obvious interest in making agentic identity sound like a solved problem, and the press release buries the real constraint, which is that EMA only covers apps that have implemented the standard. The Access Gateway fills the gap for legacy and on-premises systems, but the governance fidelity there is different and the audit trail integration is worth scrutinizing before assuming parity. The indicator to watch is how fast application vendors adopt EMA natively. If Anthropic’s launch pulls Salesforce, ServiceNow, and Workday into the standard within twelve months, this architecture becomes load-bearing infrastructure. If adoption stalls at the early adopters, the Gateway patch becomes the permanent state, and that’s a weaker security posture than the clean narrative here suggests.
Concept deep-dive: Short-lived scoped tokens
A scoped token is a time-limited permission slip that says exactly what a system, or agent, can do and for how long. Think of it like a contractor badge that opens only the server room, expires at 5pm, and can be deactivated from the front desk without changing any locks. Traditional API keys are permanent master keys. Short-lived scoped tokens limit the damage window if one is stolen, which is why they’re the foundation of modern zero-trust security architecture.
Based on reporting from C1 Announces Enterprise-Managed Authorization Support to, originally published 2026-06-18 03:00:00.
