Dylan Di Filippo
Dylan Di FilippoAI journal

Compliance API exposes saved personal instructions

AnthropicAdministration

Compliance API exposes saved personal instructions

Authorized administrators can retrieve members’ preferences in Claude and Cowork.

Enterprise beta

Official Anthropic logo — brand identifier, not announcement evidence
Brand logo. Anthropic — logo officiel ; marque de son propriétaire · Image source ↗

Verified facts

  • The `GET /v1/compliance/organizations/{org_uuid}/user_preferences` endpoint returns Instructions for Claude and Global instructions.
  • The existing Compliance Access Key must have the `read:compliance_user_data` scope; the beta is restricted to Claude Enterprise organizations.

What changes

Access scope and export retention need governance.

Practical implication

Restrict keys and retain only necessary exports.

Availability

Belgium/EEA: the source specifies no explicit regional exclusion; availability depends on the account, plan and service region.

Limits and uncertainty

This surface exposes saved instructions, not all private content or a blanket authorization for monitoring.

← Back to this weekly edition

A journal by Dylan Di Filippo. Facts, vendor claims and editorial analysis are kept separate. Coverage is not exhaustive.

Sources and methodology