Datasheet (PDF)
Monitor and Classify Sensitive Data Shared with Anthropic Claude Enterprise
When employees use Anthropic Claude Enterprise, they send prompts, upload files, and work across projects that frequently contain regulated, confidential, or proprietary data. Without direct access to Anthropic's Compliance API, security teams have no visibility into what data enters or exits the workspace. This documentation covers how Cyberhaven's Anthropic Claude Enterprise Cloud Connector integrates with the Compliance API to deliver continuous visibility into AI data exposure across chats, file uploads, and project content.
Inside This Documentation:
- How the Anthropic Claude Enterprise Cloud Connector uses Anthropic's Compliance API to monitor user prompts, assistant responses, and file uploads across chats and projects.
- What Cyberhaven inspects for sensitive data: full chat transcripts, chat file attachments, and Claude-generated files such as code interpreter output and rendered HTML.
- Step-by-step deployment and configuration instructions, including Organization ID setup, Compliance API key scopes, user and project scope controls, and scan configuration for historical and forward scans.
- Data retention constraints enforced by Anthropic, including 6-year Activity Feed retention, immediate and permanent deletion through the Compliance API, and the 25 MB content inspection file size limit, and how the connector handles those limits automatically.
Every employee Claude session is a potential data event. This connector gives your security team the visibility to see exactly what sensitive data enters your AI environment.
