Very cool project, thank you for sharing! To me, it raises some interesting questions around attribution of sources in derived works, in the same way that AI training does.