LLM-generated text/code over a certain size should add various hints that it is LLM generated. Each word count again, so you are forced to read the stuff you generate, if you want to covertly offload work (think newsapers, coders) to LLM.

And Gen-AI should add a respective watermark and media tag.