Learn how to work around AI safety filters when generating thumbnails with faces
Sometimes the AI provider's safety filters block thumbnail generation requests, especially when using inspiration images that contain recognizable faces. This guide explains why it happens and how to work around it.
The AI providers (such as Google) include built-in safety filters that can block requests containing certain faces or content. This is a provider-level policy, not a bug in Video Toolbox.
The error typically occurs when:
The most reliable workaround is to pixelate faces in your inspiration images before using them as style references.
Use a free online tool to pixelate the face:
Add clear instructions telling the AI to replace the pixelated area with a clear face:
Example: "Replace the pixelated face with Sarah, clear face, no pixels"
This tells the AI to generate a clean, unobstructed face using your selected person rather than preserving the pixelation.
Click Generate. The request should now succeed since the safety filter no longer detects the original face.
This approach works for any face that triggers safety filters:
By pixelating the original faces and instructing the AI to replace them with your selected person, you maintain the style and composition of the inspiration while avoiding policy blocks.
