In particular, only human- created content is currently eligible for copyright protection.
About a decade ago, there was a case over who owned the copyright of a bunch of selfies taken by macaques with a camera left lying out by a wildlife photographer. The US Copyright Office ultimately decided the images were public domain since they weren’t created by a human.
Because of that, AI art isn’t eligible for copyright protections.
If you make a picture book using stable diffusion and chatgpt, the only thing you can protect is the layout you did by hand of the public domain text and images on the page. Someone could sell a competing derivative work with their own original layout.