That’s the crux of the issue. But I guess only time will tell. There is some element of human creativity involved: someone needs to come up with the idea, write the prompts, select images at the very least. It can be more involved than that though: you might go through multiple iterations, select image as a starting point (either in soft fuzzy way in Midjourney or quite influential way in img2img), retouch images (to have consistent style), upscale them, categorize, describe, prepare marketing materials, etc.
It’s a new kind of thing so it’s hard to exactly know what laws will tell… the closest analogy comes to my mind is being art director. He has a big influence on the final art piece, so it probably is a joint work. So in that case it could be said that it’s a joint work between prompter and AI, but AI cannot hold copyright, so all rights would be assigned to prompter, right? But idk, this is just a speculation, we need an opinion of experienced IP lawyer really…