Microsoft announced yesterday that Bing Chat, based on GPT-4, now integrates the Bing Image Creator feature. The feature is based on DALL-E, another deep learning model from OpenAI, and can automatically generate images based on text content entered by users.
In a tweet, user Shane Parr said that the Bing Image Creator feature is very restrictive and cannot create images about “Bing” keywords.
Mikhail Parakhin, Microsoft’s head of advertising and web services, admitted that “the content warning settings are too harsh and block a lot of keywords excessively”.
Parakhin’s follow-up tweet as follows: “Friends, Bing Image Creater is too restrictive for sensitive queries, and this is a stage in the release process that we must go through. We will continue to collect feedback from our users and gradually solve these problems. Maybe tomorrow you’ll be able to use these keywords to create images in Bing Chat”.