I think this is a becoming repeating pattern in creative industries where a machine learning model is taught from a corpus on existing, copyrightable and attributable work - it creates content that plagiarizes and infringes on the rights of original authors.

I'm using "pattern" quite loosely here as recent ruckuses come from the StableDiffusion image synthesis and CoPilot - n=2.

I have no idea what the ethical and legal long term ramifications are.