Skip to main content
A workspace runs in one of two privacy modes. Protected is the default: media is kept for the retention period, and the face blocklist and duplicate detection work. Private keeps as little as it can about the person.

What it is for

Private mode is for Facial age estimation workspaces only. It also covers the ID document a person adds when the selfie cannot confirm their age. Age verification by ID document and Identity (KYC) always run in Protected mode. Moving a private workspace to the ID document method switches it back to Protected, with both options below on. Moving it back to facial age estimation does not restore Private.

What is always true in Private

No face template is kept. So: In Private mode Allow duplicate accounts is always on, since there is no face template to compare.

The two options

Both are on by default.

Where to set it

  • In the console: when you create or edit the workspace.
  • Through the MCP server: create-workspace and update-workspace take privacy_mode, private_discard_media and private_skip_device_signals; the values they accept are in the MCP tools reference. Administrators and developers can use them. On a live workspace the change is previewed first and applied only when you confirm. Private is refused on a workspace that is not facial age estimation, and the two options are ignored on a Protected workspace.
  • Not through the REST API.
The setting is fixed on each verification when it is created. Changing the workspace later never affects a verification that already exists.

What your integration and your console see

Always in Private

  • duplicate_check.checked is false on the verification.
  • POST /v1/verifications/{verification}/blocked-face answers 422 VALIDATION_ERROR, with the reason in error.message. An account without the face blocklist gets 403 FACE_BLOCKLIST_DISABLED first.

When media is deleted (first option on)

  • GET /v1/verifications/{verification}/media/{media} answers 404 MEDIA_NOT_FOUND. In GET /v1/verifications/{verification}/document, each media item keeps its id and type and has url: null. This holds at every status from creation, not only after the decision, and also in a test workspace while the verification waits for its test outcome.
  • When the person added an ID document, its fields are deleted a few seconds after the final status. Save them from the decision webhook. Automatic retries of that webhook still carry them; a webhook subscription gets them only with include_document_data: true. After that, GET /v1/verifications/{verification}/document and a webhook retried or resent by hand return the fields as null, with type and issuing_country kept.
  • The verification’s erasure field stays null, because this is not an erasure.
  • The console and the MCP server show no media for the verification.

When device data is not collected (second option on)

  • The webhook has no fingerprint_signals.
  • The console and the MCP server show no technical information for the verification.

What is kept in both modes

The decision, the status, the reason, the dates and the age result (the age and whether it passed) stay until you erase them, so you can still show why a person was approved or declined.