Avatar and Voiceover Disclosure Review (QA)

Avatar and Voiceover Disclosure Review (QA)

1. 01

   Confirm the client contract states whether synthetic presenters and cloned voices are permitted in deliverables

   Some regulated clients (financial services, healthcare) prohibit synthetic talent in customer-facing video. Get the answer in writing before a single avatar render is queued.

2. 02

   Log every synthetic element used in the cut: avatar identity, voice model, and any face-swap or lip-sync pass

   A single 60-second explainer built in AKOOL or Aistudios can stack three or four generated layers. Track them in a shared sheet so the reviewer knows what to inspect.

3. 03

   Check the platform's commercial use and likeness rights for the specific avatar or voice selected

   Stock avatars and cloned voices carry different licensing terms. A voice clone of a real employee needs a signed release on file before the video ships.

4. 04

   Verify the final export carries a provenance signal where the client's market requires one

   Google's SynthID detector went public in October 2026 with 180 billion assets already watermarked. Clients in media, education, and government procurement increasingly ask for this.

5. 05

   Screen the cut for uncanny artifacts at 1x speed on a phone before client review

   Lip-sync drift and eye-line errors read as sloppy on a 6-inch screen even when they pass on a desktop monitor. Fix or re-render before the client sees it.

6. 06

   Record the disclosure decision in the delivery notes, including what was disclosed and to whom

   If the client chose not to label AI-generated segments, note that the decision was theirs. This protects the agency if a platform or regulator later flags the asset.