Note that in this document the copyright office addresses the case of image generation, where a person were to throw prompt after prompt at the model until they get something they like, and then possibly add some more prompting to make some adjustments to get the final image. And they plainly state that this is not copyrightable in the US. Isn't this damning to vibe-coding?
[1] https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf