Feeding user data to an LLM.

I mean, somewhere, even if not now, if the Threadiverse becomes large enough, someone is gonna be using comments here to train an LLM too. That's just gonna be a given unless you want to use non-public forums, and that kills the searchability and accessbility to everyone that makes most forums valuable. It's even easier to access here than on Reddit -- just set up an instance and federate with and subscribe to everything.

I guarantee you that people are going back and training LLMs on archives of old Usenet discussions too.

The rest of it, yeah, I get.