Skip to content
English
  • There are no suggestions because the search field is empty.

Rejected by the content filter

Our AI guidelines: why the content filter stops a creation and how to create your video anyway.

Overview

Omnora creates speaker texts, voices and avatars with AI services. These services automatically check every input and every result for harmful or inappropriate content. If that check is triggered, the creation stops.

You then see the message Creation rejected due to the content filter on the video's details page. The guidelines link in the message leads to this article.

What the filter checks

Omnora creates speaker texts, voices and standard avatars with Microsoft Azure. Premium avatars are then made from this Azure voice. So the check is always done by Microsoft's content filter, in these categories:

  • Hate: discriminatory or derogatory statements
  • Sexual: explicit sexual content
  • Violence: content that describes or glorifies violence
  • Self-harm: content that promotes self-injury

Why normal content gets blocked too

The filter rates words and sentences automatically and doesn't know the business context. That's why it sometimes hits perfectly normal business content. An IT text about shutting down a database can be read as violence, a health text about illnesses as self-harm.

Step 1: Find the passage

Open the video in the editor and read through the speaker texts of the scenes. Look for words that sound drastic out of context, for example around violence, weapons, injury or death.

Step 2: Adjust the text

  1. Rephrase: replace the words with neutral terms or add context, for example “shut down the server” instead of “kill the server”.
  2. Split up: spread very long texts across several scenes.
  3. Then start the creation again, in the editor for example with Update video.

💡 For videos made from documents or slides, the passage is often already in the source material. Adjust the text there and create the video again.

What happens to your data

  • Your content is not used to train AI models. Omnora trains no models itself, and the AI providers are contractually bound not to use inputs and results for training.
  • The AI providers keep inputs and results no longer than the processing requires.
  • The AI services for text and voice run in EU regions of Microsoft Azure. The only exception is the optional premium avatar feature. It is processed in the USA under the EU standard contractual clauses, and only if your organization has activated it.

Frequently asked questions

Is this an Omnora filter?
No. The check is done by the AI services Omnora uses. Omnora doesn't decide which words are blocked.

Can the filter be adjusted for our organization?
No. The filter can't be adjusted for individual organizations. If rejections keep happening although the content is harmless, write to us with an example. We'll help you find the passage.

I can't find a suspicious passage. What now?
Write to us in the chat and give the title of the video. We'll check which scene is affected.

Questions? Send us a message, we are happy to help!