A good reason for courts to find that training MOLE(1) on published works is copyright violation, but also a reason they probably won't;
"Given the unimaginable reams of copyrighted works that have presumably already been incorporated into systems like GPT-4 and Claude, you don’t even need to do the math. AI companies, even gigantic ones, could be bankrupted by the damages they owe to—well, all of us."
#MattPrewitt, 2024
https://www.radicalxchange.org/media/blog/the-models-are-yours-the-publics-leverage-in-ai/
(1) Machine Only Learning Emulators
(1/?)