Sta­ble Dif­fu­sion con­tains unau­tho­rized copies of mil­lions—and pos­si­bly bil­lions—of copy­righted images.

if what I remember from my AI class is correct, this is not true at all. While the training data may have contained these images, the AI itself does not contain these images. All the AI file is is a collection of nodes with different number values assigned to them. they can data mine the AI files all they want, they are not going to find any images of theirs. Additionally, as far as I'm aware, stability AI never distributed their training data.

As for collecting the training data, assuming the training data was not full of pirated images, then what people do with the shit that you post on the internet is none of your business.