Hi Google AI team,
We recently got allowlist access approved for gemini-omni-flash-preview on project Skyhill HQ (project ID: skyhill-hq, project number: 7000176789..) - thank you for that.
However, every generation request now fails with:
“Input blocked: Sorry, we can’t create videos with real people’s names or likenesses. Please remove the reference and try again.”
This happens even after removing all explicit text references to a person’s identity from the prompt (we only describe generic editing instructions like camera motion, background music, and text captions). The source video is our own founder’s brand/marketing footage that we have full rights to use and edit - we are only trying to add camera motion, background audio, and kinetic text captions on top of the existing footage, not regenerate or alter anyone’s face or identity.
Is there a way to get this reviewed/appealed for legitimate first-party footage use cases (own brand content, rights-cleared), or is this a hard, non-appealable restriction for any video containing a real human face? If there’s a required disclosure, consent attestation, or different workflow (e.g. Vertex AI vs AI Studio) for this use case, guidance would be appreciated.
Thank you!