The part that the human created, the prompt, can be copyrighted.
The part that the LLM created, cannot be.
Copyright in code works exactly the same way: the source code is copyrighted. The binary code is only copyrighted to the extent that it is derived from the source code. This is well-established.