I've made these LLMs say all kinds of inflammatory shit. There isn't a religion they won't offend with just a few prompts. TBF I'm only using offline LLMs, but they are just quantized versions of the commercial ones. It's like working with low res images because your computer can't handle high res. The fundamental "information" is the same. I think they are useful in the sense that you can blame the AI for saying shit it would not be safe for you yourself to say.