Reddit probably was a pretty big part of LLM training data. It has a vast trove of conversation on niche topics, and pairings of "what is said in response to input X". Probably the monetisation of the API is an attempt to capitalize on those users.
Yea, the future of public discussions online is interesting to consider. LLM bots are already a thing, comments posted here could be from a bot. We don't have a way to really confirm it, and I am not sure it actually matters (yet).