At some point this is going to start bumping up against things like procedurally generated assets and adding filters or rendering textures in images. There’s some point between adding programmed randomness to art and bringing an image back and forth between AI generation and adjustments on the one hand and traditional techniques on the other that our line has wider implications.

Personally, I suspect it probably has more to do with how the person submitting the copyright frames the issue than with any real line in the technology. Nobody’s getting bent out of shape about shitty rendered Photoshop clouds because they aren’t making the same claims.

It would be interesting to see someone try to register a copyright for an image created using fairly common filters and rendering techniques and frame it as AI generation obfuscated by specific technical technology that brings the idea to mind without actually misrepresenting it as stable diffusion or whatever.