I wonder if we’ll eventually see an open training set composed entirely of AI generated materials by OAI and others’ models. This API restriction is only an is

I wonder if we’ll eventually see an open training set composed entirely of AI generated materials by OAI and others’ models. This API restriction is only an issue until that point, I assume, or is the distilling being done in these situations much more dynamic in nature?