Generation · troubleshooting
Why your AI video prompt was rejected
An AI video prompt is rejected when a content filter reads it, or the image you uploaded, as violent, sexual, or about a real person without consent. On Percify prompts and uploads are checked, and most model providers run their own filter on top, so a rejection can come from either. Legitimate prompts are mostly flagged for wording that has a violent or explicit reading out of context; say what you mean literally and they go through. A generation that fails is refunded automatically.
- 2
- layers of checks: ours, then the model provider
- Auto
- refund when a generation fails
- Violent
- the category named in our own rejections
- 1
- plain rewrite, usually enough
What gets a prompt rejected
Filters look for a short list of things, in the prompt and in any image you upload:
- Violence or gore, including innocent words that read as violent out of context.
- Sexual or suggestive content.
- Real, identifiable people who have not agreed to it, public figures included.
- Minors in any sensitive context.
- Uploaded images that fail the same checks.
Why legitimate prompts get flagged
Filters read words, not intentions. "Shoot the product on a table", "a killer hook" and "explosive growth" all mean something harmless and all contain a word with a violent reading. The fix is to say what you mean literally: "film the product on a table", "a strong hook", "fast growth".
That is clarity, not a workaround. If what you are asking for is genuinely violent or explicit, rephrasing it is not the answer, and it will not be generated here.
Where the rejection came from
The message tells you which layer stopped it:
| Message you see | Where it came from | What to do |
|---|---|---|
| Prompt rejected by content moderation | Percify's check on your prompt | Rewrite the flagged idea literally |
| Input was rejected by content moderation | The check on your input or upload | Use different wording or a different image |
| The input or output was flagged as sensitive | The model provider's own filter | Rephrase, or try another model |
Real people and uploaded photos
Only use faces you have the right to use: your own, or someone who has agreed. Public figures are off limits even where a tool will technically accept them. If you want a person on screen who is not you, use a preset avatar or a generated character. The reasoning is in ethical AI video generation, and your own likeness is set up in clone yourself.
What happens to your credits
A generation that fails is refunded automatically, so a rejection does not cost you the run. If a generation completes but looks wrong, that is a different problem; see fix AI video glitches.
Get a legitimate prompt through in three steps
- 1
Read the message
It tells you whether our check or the model provider stopped it.
- 2
Rewrite the idea literally
Replace words with a violent or explicit reading with what you actually mean.
- 3
Swap the image if it was the upload
Use a photo you have the right to use, without the flagged content.
What goes through
Real Percify output: people, products and characters, described plainly.
Product ad
a product, described literally
Alex
a person who agreed, from one photo
Tyler
a generated character
Frequently asked questions
Why was my AI video prompt rejected?
A content filter read the prompt, or an uploaded image, as violent, sexual, or about a real person without consent. On Percify, prompts and uploads are checked, and many model providers run their own filter too.
How do I fix a prompt flagged as violent?
If the idea is harmless, say it literally. "Film the product" instead of "shoot the product", "a strong hook" instead of "a killer hook". If the idea is genuinely violent, it will not be generated.
Can I make an AI video of a celebrity?
No. Real, identifiable people need to have agreed, and public figures are no exception. Use yourself, a preset avatar or a generated character.
Do I lose credits when a generation is rejected?
A generation that fails is refunded automatically, so a rejection does not cost you the run.
Why was my uploaded photo rejected?
Uploads go through the same checks as prompts. Use a photo you have the right to use, without explicit or violent content.
Keep reading
Say what you mean
Plain words, faces you have the right to use, and the run goes through.