Anthropic adds user profiles to Claude API for per-customer enforcement

Anthropic shipped user profiles in the Claude API, letting resellers tag each request with an anthropic-user-profile-id header so a ban hits one end customer instead of the whole organization. On October 21 it added enforcement_status and enforcement_actions fields, but the organization webhook that should report bans accepts registrations and delivers nothing.

• Profiles are not billable and do not affect pricing
• Profile creation is not idempotent: every call mints a new uprof_ ID
• Bedrock and Google Cloud validate only the ID format, not the profile
• Organization webhooks accept registrations but send no events yet