I was toying with text-DaVinci-003 model and it seems while it can easily produce poems about specific topics, it can not produce bible/quran verses. Instead it produces actual verses from these books which are quite irrelevant to the topic I asked for.

I was wondering if this was hard-coded by OpenAI (99.9% this is the case), or the model just naturally generates text from original books because they are so prevalent on the internet.

And if OpenAI did this, why? I don't remember anybody complaining about the models being able to generate these texts.