Share via

Why any prompt related to Hindu god is set as violence in the DALL-E playground?

manisha c 0 Reputation points
2024-08-08T05:04:14.2533333+00:00

I had entered the prompt in the DALL-E playground for creating the image. The AI has thrown this error "The prompt was filtered due to triggering Azure OpenAI’s content filtering system. Reason: This prompt contains content flagged as Violence (low)".

The prompt I wrote was " "Indian god Ram with his wife Sita and brother Laxman in a beautiful jungle, standing by a river. Ram needs to be in bluish skin tone".

I want an explanation on how and why this prompt is flagged as violence. Please help me understand.

Bing Image Search
Bing Image Search
A Bing service that supports searching for images and gets comprehensive results.
48 questions
Azure OpenAI Service
Azure OpenAI Service
An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
3,118 questions
0 comments

2 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. dupammi 8,465 Reputation points Microsoft Vendor
    2024-08-08T06:15:38.3733333+00:00

    Hi @manisha c

    Thank you for your question.

    By default, content filters are enabled and set to a medium severity level. To gain full control over content filtering, including setting filters to high severity or turning them off, approval is required.

    Only managed customers/partners can opt-in for this. See How to configure content filters with Azure OpenAI Service.

    I hope you understand. Thank you.

    Please don’t forget to Accept Answer and Yes for "was this answer helpful" wherever the information provided helps you, this can be beneficial to other community members.

  2. Suzana Ilic 0 Reputation points Microsoft Employee
    2024-08-09T14:07:19.4366667+00:00

    Thank you for flagging this! It's likely a false positive, our team will look into it. The default policy for DALL-E is set to severity threshold low for content harms categories such as violence. You can read more about default settings and what different severity levels mean and entail here https://zcusa.951200.xyz/en-us/azure/ai-services/openai/concepts/default-safety-policies

    You can create a custom content filtering configuration and change the severity threshold to medium if needed, and assign it to your DALL-E deployment. You can find instructions here https://zcusa.951200.xyz/en-us/azure/ai-services/openai/how-to/content-filters

    Hope this helps!

    0 comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.