
The rules live in the Terms of Service, section 04. This page explains how they play out while you work.
The Terms, section 04, prevail; in everyday words: illegal content — sexual or pornographic material, terrorism and extreme violence, gambling, drugs, and anything else the law forbids; harm to real people — defamation, privacy violations, and using someone’s likeness or voice without consent (uploading their photo as a character counts); unauthorised IP — copyrighted film, anime and game characters and footage; the provider runs an “IP suspected” check, and celebrity faces, well-known brands and characters trigger it; evasion and tampering — getting around moderation with respellings or code words, removing or altering the AI-generated label, scraping, and using multiple accounts to dodge a sanction. Any sexualisation of minors is zero tolerance regardless of context.
Moderation runs at the model providers, on every generation. For character, scene and prop sheets your description text is checked first, then the image; a block usually names a word in the description. For video, the prompt, the reference images and the clip are checked; the card says the content was blocked by the safety rules, and hovering it shows the provider’s original message, which may name an IP check. Parsing, AI Create and translation check the text. And people look too: feedback, complaints and spot checks. The checks are imperfect — ordinary fight scenes, medical or historical material is occasionally blocked and passes after a rewording. One block is not a finding against you; repeated blocks are the signal.

This attempt: refunded automatically when the provider confirms the block, up to 12 blocks in 24 hours; held for review past that or when the outcome is uncertain — see Refunds. Repeated in a short time: more than 12 blocks within 24 hours pause video generation for 1 hour; the count does not reset inside those 24 hours, so each further block pauses again. This stops moderation being used as a lottery. Parsing, sheets and storyboards keep working during a pause. Serious or deliberate: content is removed, features are restricted, the account is suspended (a balance left on it is handled under the Terms, sections 03 and 09), and we report to the authorities where the law requires it. Credits and membership left on a suspended account are not refunded. That decision is made by a person looking at the generation history, never on a single false block. Real people in uploaded photos: the ad flow removes people from a photo before using it; character and scene photos must be material you have the rights to, and substantiated complaints are handled under the Terms.
Read whether the message names the text or the image. For text, edit the description: remove explicit, gory or real-person terms, brand names and IP characters, and replace “result words” with neutral description of what the camera sees. For an image, the description usually caused it — edit the description and regenerate. For a blocked video, change that segment’s storyboard text or the sheets it uses; resubmitting unchanged only gets blocked again and counts again. For “IP suspected”, replace celebrity faces, known characters and brand marks with original designs; realistic styles hit celebrity likeness more often, so change the style or the appearance text. Subjects that cannot be reworded — heavy gore, sexual content, real political figures — are not made here. Step-by-step rewrites: Blocked by Moderation: What to Change.
Help → Feedback, category Other, with the task number, what you wrote and the message you saw; a person reviews it. A false block gets its held credits returned; the pause itself simply runs out. A confirmed violation is handled as above and you are told why. A suspended account cannot use the site, so a suspension appeal goes by email to support from the account’s email address. How to get help.
| Where | What is checked | What you see when blocked |
|---|---|---|
| Character, scene and prop sheets | Description text first, then the image | text check failed / image check failed |
| Video generation | Prompt, reference images, the clip | Blocked by content safety rules; hover for the provider’s message |
| Parsing, AI Create, translation | Text | input contains sensitive content |
| People | Feedback, complaints, spot checks | A notification or an email |
No. Occasional blocks in normal work are common. Only repeated blocks in a short time pause generation, and only deliberate violations are sanctioned.
Yes. Only video generation pauses; parsing, sheets and storyboards continue.
Yourself, or someone who gave you permission. Not someone else’s face.
Not without authorisation. The provider blocks it and repeated attempts are sanctioned.
A balance left on a suspended account is handled under the Terms, sections 03 and 09.
Members can switch off the visible badge on their own clips. Removing or altering AI marking yourself is not allowed.
Everything the checks block has a workable alternative: an invented face, a fictional brand, a scene described rather than shown.
Updated September 27, 2026
Pricing·Novel to Video·Guides·Samples·Support·Privacy·Terms·Legal·Contact·© 2026 SceneMixer