Gentoo Linux's AI policy forbids any content, including code, created with LLMs, for contributions to official Gentoo projects due to copyright, quality, and ethical concerns.

https://wiki.gentoo.org/wiki/Project:Council/AI_policy

In the age of batshit AI companies like Google, Microsoft, OpenAI, and others, a few opensource projects are making the correct call. Can Linux foundation also ban LLM?

#AI #LLM

@nixCraft I have one sincere question: is the correct step tô ban LLM or is the correct step tô tighten reviews so that LLMs can be applied às another tool for development?

We had bad contributions before LLMs.

What if the contributions don’t mention LLMs and are accepted? Will they be reverted?

@nixCraft i like it but i doubt they can enforce it

@nixCraft Well done.

@nixCraft if that policy is fully implemented, meaning that all AI traces will and can be detected, there will be little software left.

It's hard to notice if e.g. a one-liner is AI generated, but thx to copilot it often is. What about AI in analysis? What in doc? What about lazy human patches?

I think the same safeguards should be applied to patches no matter what the origin is. Are eg. US citizen patches better than from china or russia? Or is it just matter of quality?

@nixCraft that's a really good, dry delivery there

@nixCraft The LF has essentially left the LLM question to each of its sub-projects to answer; official guidance at:

https://www.linuxfoundation.org/legal/generative-ai

@nixCraft If there's something I learned from contributing to open source projects is that under the scrutinity of a patch review, you should be able to explain why every line written in a patch is there.

I seriously doubt a person contributing a patch generated by AI fully understands what the patch does, let alone what every line does and why is there. Even worse, many open source projects have to face patches generated by AI where the patch doesn't even apply. It's a time waste. Thus the ban