I don't get this whole "your upstream is using AI, and so should you! just give in!" thing is coming from. This isn't a race to the bottom, people.

Some downstreams just accept the inevitable: someone somewhere upstream will do something at any some arbitrary time to do something unwise.

Decisions made upstream are not immediately inherited downstream (guess why the downstream exists already.)

Downstreams often make different decisions than their upstreams. And that's totally fine!

As the maintainer of multiple open source projects with upstreams: I cannot force my will on others. I myself won't use AI/LLM in this context. I will reject patches whose authors admit to using AIs/LLMs in the development of the patch. (I recognize that's based on "the honor code"--again, I cannot force people; I can only lead by example.)

I will primarily reject AI/LLM generated patches because I do not have the resources to pay a lawyer. As the maintainer downstream, there is legal risk in accepting (knowingly) generated code due to the large potential for copyright infringement or other intellectual property/licensing rights.

In one project, I have strictly maintained separation from another codebase in order to perform a clean-room reimplementation. After having spent more than a decade strictly trying to avoid license taint, why would I want to risk it now?

I think you'll find a lot of wisdom in those people who remain steady while others get easily beset by the wind.

(Please know that I'm human, I make mistakes. It's possible I misphrased something here, and if so, I apologize. I did note that I am an idiot, and these are simply musings.)