This is largely to prevent the first six posts on the topic from being people fighting over it
I’m not sure it’s working. If the author of a project decides their code or writing was not “substantially LLM-generated” and thus that they should not, and do not, put the disclosure in their post…
Then certain users will still go through the code, find evidence of LLM use, and post about it here.
I understand that you do not want to put a thumb on the scale, but right now the “anti-LLM people” you mentioned are hijacking every post that does not disclose any use of LLMs, not just posts that do not disclose substantial use of LLMs.
I almost exclusively use LLMs in a “read” capacity and do not allow models to write for me (code or otherwise). If I shared one of my projects here, it would likely have an AGENTS.md file in the source tree, but none of the code or other parts of the project would have been generated by a model. Despite that, the anti-LLM people would still immediately show up and try to tell me I need to put a disclosure on my post, even though according to the rules I would not. I seriously doubt my response informing them as much would not just receive increasingly hostile replies from certain users (as has been exhibited over the past few weeks on this forum).
I think the much better solution/policy would be to not require a disclosure at all, and to quickly delete off-topic comments complaining about LLM usage on link posts. The users who are most concerned about LLM use will search for and find it anyways. There is no reason to keep giving them an “in” to derail more discussions.