I'm worried about the future of open-source

At first glance, this looks like the commit history of just another random vibe coded project. But no, it is in fact the recent commits to go-micro, a project with 22.7k stars that started all the way back in 2015. BcacheFS, a major filesystem, now has ProofOfConcept (a custom AI bot) as co-author on the majority of new commits. The Claude account on GitHub now has so many commits that its GitHub profile fails to even load.

This worries me.

When it's Claude code randomly creating 10GB VM bundles, we point and laugh. When Cline gets supply chain attacked because of a prompt injection, nobody cares because dependence on them is purely individual. Very few build on top of those shaky towers.

But now AI is coming for the foundations of what we build. I personally don't have much confidence in either the quality or security of what AI pumps out.

chardet recently got completely rewritten by AI as a means of license washing (getting rid of LGPL) while breaking various behavior and generally making the quality worse. Again, not a small package. It is a dependency of almost 1 million packages tracked by GitHub.

I don't even know what I want to say yet, other than that I do not like this new iteration of open-source and I believe this behavior will become more widespread over time. Barley anyone gets paid for it, and LLMs are an easy way out of spending the effort maintaining a library.