Anthropic adds a security center and generally available inference hooks for Claude Enterprise

Anthropic's page on new security and admin controls for Claude Enterprise has a publication date of 5 November 2026. The Claude release notes for that day describe the same controls and link to the page. The page says the controls are for Claude Enterprise, with some exceptions during beta listed in the Help Center. That exceptions list was not opened.

The page says a security center, in beta, lets a team review settings for chat, Cowork and Claude Code on one page and compare them with Anthropic's recommended configuration. It says the page can be downloaded as a report. It says the center is guidance based on Anthropic's recommendations, not a certification or a guarantee of compliance with any law or standard. It says organizations with the Compliance API can read the same results through that API.

The page says inference hooks are generally available. Prompt hooks send prompts from supported Claude products to the organization's security tool for a check. Pre-tool-call hooks check tool calls in supported products before they run. The page says centralized management, also in beta, gives a parent organization one admin console at admin.claude.com for single sign-on, directory sync, domain verification, groups and Compliance API keys across its Claude Enterprise and Claude Platform organizations. It says setup cannot be undone.

The page says signing a member out everywhere through the Admin API, and a self-serve IP allowlist, are in beta. It says admins can block personal and other outside Claude accounts in Claude Desktop on managed computers. It says an endpoint-detection sensor inside the Cowork virtual machine is in beta on macOS and Windows, for Claude Team and Enterprise and for the Claude desktop app on Amazon Bedrock, Google Cloud Vertex AI and Microsoft Foundry. It says the sensors run in detection mode: they report activity and do not block it, and the coverage is the virtual machine on the user's computer, not Cowork sessions in the cloud. The page says supported vendors are listed in the Help Center. That list was not opened, so no vendor is named here.

What it means for business: inference hooks are the part the page marks as generally available. The security center, the parent console, the sign-out call, the IP allowlist and the endpoint sensor are marked beta. The page says the security center is not a certification. This describes the page. It is not advice on a control setting.