Doesn’t it usually have to be queried with a bunch of the training material anyway? It’s a pretty foolish sounding argument that one is being harmed if it boils

Doesn’t it usually have to be queried with a bunch of the training material anyway? It’s a pretty foolish sounding argument that one is being harmed if it boils down to “when I type in the first half of my article into GPT it can sometimes complete a few more paragraphs of it” (That’s what the NYT tried to argue/show. I’m not sure if that case is ongoing or settled.)