Anthropic adds an Enterprise security center and generally available inference hooks

On Nov. 5, 2026 Anthropic said Claude Enterprise has a new security center, still in beta, that compares chat, Cowork, and Claude Code settings with the company's recommended configuration. The same post says prompt hooks and pre-tool-call hooks are generally available, so an organization's server can allow or deny a prompt or tool call before it runs. Centralized management, member sign-out, and the self-serve IP allowlist are in beta, and the docs say the hooks do not cover Amazon Bedrock or Google Cloud.