Calling it copyright today is a misnomer, it's not actually the act of copying the work that's a problem; it should actually be called "Performance Rights" or "Redistribution Rights." The part where this gets complicated is that OpenAI has (presumably, if they haven't that's a different matter) acquired the works through legal means. And having acquired them they're free to do most anything with them so long as they don't redistribute or perform the works.

The big question is where "training an AI on this corpus of works and then either distributing the weights or performing the work via API" fall? Should the weights be considered derivative works? I personally don't think so and although the weights can be used to produce obviously infringing works I don't think this meets the bar of being a redistribution of the work via a funny lossy compression algo like some are claiming. But who knows? Copyright is more political than logical so I think the bend is really gonna be a balance of the tangible IRL harms artists can demonstrate vs. the desires of unrelated industries who wish to leverage this technology and are better for having all this data available.