CineMake.ai is an Agentic AI platform used for video generation.
Please visit my app’s splash page for more information and a demonstration of my product:
cinemake
We are experiencing frequent false-positive blocks from the Celebrity safety filter (support codes 29310472 and 15236754–mostly 15236754) across both projects(preprod and prod) in our usage of Google Veo through the Vertex API. The filter is rejecting legitimate generation requests for original, non-celebrity character avatars and stylized human figures that our users create for storytelling projects. These avatars are 100% original designs generated or uploaded by our users — no real celebrities or public figures are ever intended or allowed in our application.
we are getting many false positives like descriptors below are failing. We need whitelisting because this is affecting the quality of our app. We do not allow celebrity likeness but are getting many false positives. An example of something that leads to a false positive below:
Attaching this descriptor for anchoring characters and anything similar is giving us false postives:
woman, pastel hair, bold brows, trendy streetwear, confident smile, energetic posture
Our Compliance Position
• We strictly enforce Google’s Generative AI Prohibited Use Policy and Acceptable Use Policy.
• Our app does not allow or generate celebrity likenesses.
• We have built-in guardrails and content moderation to prevent any prohibited use.
• We fully align with Google’s Responsible AI Principles and Veo’s safety guidelines (as documented at https://cloud.google.com/vertex-ai/generative-ai/docs/video/responsible-ai-and-usage-guidelines).
Business Need
Our core product relies on Veo to generate consistent character avatars across video scenes. The current false-positive rate on both projects is severely impacting user experience and product functionality. Adding both projects to the allowlist will allow us to continue building safe, creative tools while remaining fully compliant.