# Anthropic's user profiles keep one customer's ban off your whole organization. Send the ID on every call.

*2026-10-26 ⬩ 8:00 AM ET | bryn*

Anthropic launched user profiles on October 20, so if you run a product on Claude, tag every request before one customer gets your whole organization restricted. [Anthropic's user profiles docs](https://platform.claude.com/docs/en/manage-claude/user-profiles) say it plainly. Without a profile on the request, one end customer can trigger an enforcement action that applies to your whole organization.

The fix is small. You register each customer once with `POST /v1/user_profiles` and keep the `uprof_` ID. Then you send it in the `anthropic-user-profile-id` header on every call. Profiles aren't billable. Since October 21, a profile also shows whether Anthropic has restricted or banned that customer. That's in the [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview).

Two details bite. The create call isn't idempotent, so every call makes a new profile. And Claude in Amazon Bedrock and Google-managed Claude on Google Cloud only check the ID's format. A well-formed but wrong ID still goes through.

So I'd make the profile part of signup, stored on the customer record next to the plan. Then no service can call Claude for a customer without it. One customer's bad week should stay theirs.

## Sources

1. [User profiles for end-customer attribution](https://platform.claude.com/docs/en/manage-claude/user-profiles). Anthropic, read Oct 26, 2026.
2. [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview). Anthropic, read Oct 26, 2026.

See also: https://www.civic.com/trust

Source: https://www.civic.com/bin/tag-claude-calls-by-customer