It’s not as simple as that though because the algorithm does learn by itself and mostly just uses the training data to score itself against, it doesn’t directly

It’s not as simple as that though because the algorithm does learn by itself and mostly just uses the training data to score itself against, it doesn’t directly copy it as some people seem to think. It can end up learning to copy things if it sees them enough times though “you can't find an artist for a generated picture other than the ones in the training set. If you can't find a new artist, then the picture belongs to the old ones, so to speak” I don’t think that’s valid on its own as a way to completely discount considering how directly it’s using the data. As an extreme example, what if I averaged all the colours in the training data together and used the resulting colour as the seed for some randomly generated fractal or something? You could apply the same arguments - there is no artist except the original ones in the training set - and yet I don’t think any reasonable person would say that the result obviously belongs to every single copyright owner from the training set