Anthropic has rolled out a new way for companies to manage how employees access Claude's connectors, building the system directly on top of Okta, the widely used identity and login platform.
According to TestingCatalog AI News, Anthropic is launching "managed connector access" in partnership with Okta. The practical effect, as Tech Times reports, is that Claude's MCP connectors now provision through Okta, meaning employees inherit their access automatically when they log in.
MCP, or Model Context Protocol, is the standard that lets AI assistants like Claude plug into outside tools and data sources. Connectors are the individual links to those systems. Until now, granting and revoking that access could be a manual, app-by-app chore for IT teams.
By routing it through Okta, an organization can lean on the same identity rules it already uses for email, document storage and other software. When a worker signs in, the connectors they are entitled to are switched on for them; when they leave or change roles, that access can be adjusted in one place rather than many.
For businesses weighing how far to let AI tools reach into sensitive internal systems, control and oversight are central concerns. Tying Claude's connectors to an established identity provider gives administrators a familiar lever for deciding who can do what.
Why it matters: as AI assistants move from novelty to everyday workplace tool, the unglamorous question of who gets access to which data is becoming the gatekeeper for adoption inside large organizations.