Content Filtering in Azure GPT

Question

I'm currently developing a service using Azure's GPT API.

Questions that previously worked without issue are now not receiving proper answers due to content filtering.

I've lowered the content filtering level of the deployed models to the lowest level, but the issue persists.

I provide a service based on the content of documents. If a user's document contains a manual for operations, content such as permission adjustments is filtered, preventing proper responses.

Also, links within documents that are too long seem to be filtered. The current strict requirements are interfering with the normal service provision.

I requested separate approval by referring to https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/content-filters#understand-content-filter-configurability

,

but we need to quickly decide on a future direction. If this continues to be an issue, we may need to consider migrating to another API, such as Gemini.

I would appreciate it if you could give us some insight into how we can improve the situation and how long it would generally take for that to happen.