To no ones surprise at all AI art has made it to the courtroom, and although my only source on this is the below video which is also a pretty good summary of general free-use law I will share some thoughts on the topic of copyright in general.
Mostly I think if in was not Stability AI then it would have been someone else. Did stability not cover their asses well enough and get a bit sloppy with the "mainstream" fame of improving art generation specifically in the style of [insert artist]; sure they did.
Then again, just because Dall-E and Google Imagen aren't under fire does not mean they are not doing the exact same thing. The difference seems to be merely that Midjourney and Stable Diffusion are more widespread and open.
Now when it comes to copyright tears and little bitch babies whining about digital content they freely put online being used by others for whichever purpose; I have very limited sympathy.
I really think anyone first though needs to be since the dawn of the internet, do I want this to ever be used by someone else? If No, then don't.
After that it gets a bit murky, and as the guy pointed out in the video Google won a case for showing images in search results as not being copyright infringement.
Now by extension, if the AI model used searched results as their source imaging my tiny brain would go oh well we tried I guess it is all fair and square.
Obviously, it is not so simple, since I can't use Google Image results as a cover image in my posts or people get very uppity and wanna poke a splintered stick up my whooza; all in all copyright and free use is a big hoopla.
But...
If I consider current laws in place, and although I think copyright law is a load of precum it does not excuse the fact that people do infringe.
Aside from it being convoluted.
I am certain that Stability AI expected a lawsuit to come at some point. If they did not then they clearly did not do their due diligence and deserve to get sucker punched.
Now if I was an overly cautious person and realise that in general people are nothing but a mustard yellow cyst and I made a system that could push the boundaries of art as we know it but it needed as many of the worlds images as possible and knowing all of those belonged in part to some cyst and being averse to having puss all over my face and not liking the idea of jello-like ooze in my eyes; being unsure if it burns as much as it stinks.
I might look to make it distributed since then everyone else can sqwoosh (not a word) all over everyone else when it comes to something silly as a copyright claim.
You know, like telling an artist they can't draw a picture they saw on the internet and all that; or telling woman they need to allow pedos into the girls bathroom because of mental issues - I mean, identity.
I don't think the previous but... explained anything
The guy in the video is not wrong though, it will set precedence for all future use cases of digital content generated by AI.
However, as you cannot tell someone not to learn to draw pictures from the internet, or not to learn to play a song just like Beethoven by ear. I don't think they can stop the AI from doing what it does and it will fall on the end user and how the final product is used.
Artistic style transfer is a thing already and whether the current AI needs 100's of a person's images to accomplish it is merely a limitation of the software.
So possibly the case can be made that it does in fact use original samples although tweaked a bit but samples nonetheless instead of completely drawing its own version based on geometry and insight.
The difference is, a photoshopped image vs a newly drawn image from memory or reference as base.
For me the photoshopped image uses the original as a base and can then in part be said to contain copyright images.
The latter uses an idea, I don't think you can copyright ideas.
I am sure there are many more points a person can dive into but in general I am coming from a place where copyright is at best irritating, then again I am not a content creator.