This only makes sense if we have open access to the training data so can verify if it’s copyrighted or not. Otherwise how am I supposed to know it’s replicated

This only makes sense if we have open access to the training data so can verify if it’s copyrighted or not. Otherwise how am I supposed to know it’s replicated someone’s IP.