The splendidly-named "OpenSlopware" was, for a short time, a list of open source projects using LLM bots. Due to harassment, it's gone, but forks of it live on.
"OpenSlopware" was a repository on the European Codeberg git forge containing a list of free software and open source projects which use LLM-bot generated code, or integrate LLMs, or which show signs of "coding assistants" being used on the codebase, such as pull requests created or modified by automated coding tools.
However, its creator – who we are intentionally not naming or tagging here – received so much harassment from LLM boosters that they removed the repository, and indeed their Bluesky account, stating that they would withdraw from social media for a while. Now, if you try to visit the original URL, you will receive only a 404 message.
I'm always pro-information availability, so this sucks, I hope the repo was archived and returns and the FOSS paradigm proves resistant to harassment and the repo finds a new home.
I think the title of this thread is a tad inflammatory though, as for the quality of AI code, I mean, tbqh it's more an issue of maintainers and standards. I've seen plenty of code written by flesh and blood people they themselves couldn't explain either. Like if there was LLM code in the Linux kernel, I'd trust it no less than any other code in the Linux kernel because I trust Linus and others working on it and their eagle eyes and high standards for spotting problems and bugs, having passed the same standard I'd have absolutely zero issues with the code regardless of where it came from.