How to Test Content Filtering Without Triggering Abuse Monitoring in Azure OpenAI

Question

How can I test the content filtering functionality in Azure OpenAI without triggering abuse monitoring?

PS - Based on common issues that we have seen from customers and other sources, we are posting these questions to help the Azure community.

Answer 1

To test the content filtering functionality in Azure OpenAI without triggering abuse monitoring, you have a couple of options.

• Firstly, you can create a custom content filter and adjust the severity level to meet your testing needs. This allows you to control how the content is filtered based on the required criteria.
• Secondly, you can apply for modified abuse monitoring. This will enable you to tailor the abuse monitoring settings to accommodate your testing scenarios without triggering the default abuse alerts. If you are an approved customer, you will have the ability to make these adjustments. Please ensure you apply for the necessary permissions to modify these settings.

Resources:

• Azure OpenAI Service Documentation
• Content Filtering
• Configure content filters with Azure OpenAI Service
• Azure Abuse Monitoring

Please do not forget to "up-vote" wherever the information provided helps you, as this can be beneficial to other community members.