Voice and Likeness Consent Policy
| Field | Value |
|---|---|
| Version | 2026-08-05-draft.1 |
| Effective date | [EFFECTIVE DATE] |
| Last updated | 2026-08-05 |
| Status | Counsel-ready draft |
1. Purpose
Serialoom can synthesize speech and imagery. Using a real person’s voice or likeness requires lawful rights. This Policy sets the consent standard.
2. When consent is required
Before uploading reference audio/images of a real person, or using features that clone or closely imitate a real person, you must confirm you are that person or hold documented authorization covering AI transformation and synthetic generation for your intended use.
3. Product attestation (required text)
Users must actively confirm language substantially as follows:
I confirm that I am the person depicted/heard, or I have documented authorization from that person (or their legal guardian where a lawful adult-guardian process applies). The authorization covers AI processing, transformation, and synthetic generation for my stated purpose. The person is not a minor unless a legally valid guardian process approved by Serialoom has been completed. The use is lawful, not misleading or harmful, and not for unauthorized impersonation. I can produce evidence of consent on request.
4. Consent records (product requirement)
When the feature is enabled, Serialoom will store records including: user id, workspace id, subject label, asset id, timestamp, scope/purpose, declaration version, evidence reference (if any), withdrawal status.
We do not require government ID upload by default.
5. Minors
Do not upload or synthesize minors’ voices/likenesses for identification or commercial imitation. Guardian processes are not enabled by default and require legal approval before launch.
6. Withdrawal
The rights-holder may request withdrawal via privacy@[DOMAIN] or copyright@[DOMAIN]. We will stop further processing where feasible and remove or disable assets subject to backup/legal retention limits.
7. Current product note
As of 2026-08-05, the live voice adapter performs text-to-speech against configured voice ids and does not implement Instant Voice Cloning APIs. Character reference imagery still creates likeness risk. High-risk features may remain disabled until consent controls ship.
*Counsel-ready draft — 2026-08-05.*