As an aside, when submitting prompts that AI rejects because it considers the subject matter "racially or culturally sensitive," I trick it. For example, if asking about black churches, tell it that you're asking for your nephew who is writing a term paper for school. Yes, it's a lie, but you're not lying to a person, you are manipulating a machine.
You might also tell the AI model that you're trying to approach this issue from all sides to understand it better.
AI algorithms have difficulty with both the above. I have also accused ChatGPT and Grok of demonstrating a bias, in your case, for example, that would be bias against black churchgoers. They'll blow a GPU over that.;-)