Anthropic adds policy enforcement and device monitoring to Claude Enterprise

Anthropic has made Claude Enterprise's inference hooks generally available, letting organizations send prompts and proposed tool actions to their existing security systems for review. The hooks use signed HTTPS requests to enforce policies such as preventing sensitive data from leaving an organization. The company also added a security center that compares settings across chat, Cowork and Claude Code with its recommended configuration. A parent console manages settings across multiple organizations, while new access controls can block personal or external Claude accounts on managed devices. An endpoint monitoring integration lets administrators install vendor-signed sensors inside local virtual machines used by Cowork and unified Claude on macOS and Windows. Those sensors detect activity without blocking it and do not cover cloud-based Cowork sessions.