Azure Open AI GPT4 with vision causes content filtering for normal images
Hello!
We are having issues with the content filter using vision capabilities of Azure OpenAI, for some reason, our documents are getting filtered due to random categories such as violence and sexual. But sending a sample document page containing only a piece of the Lorem Ipsum makes the filter pop up. In our scenario we are extracting PDF pages and sending each page as an image to the model and it should only do OCR.
Any way around this? We tried sending the form to remove the content filter but we got declined.
Due to our app being in the Medical field it is normal that sexual topics arise.
We have already setted the filter to the highest setting, but still we see these issues on vision.