Establish a policy for code generated by LLM assistants

[emersion]

LLM assistants are more and more widely used by contributors, and may raise issues related to e.g. copyright and code review. It would be nice to decide on a policy for their use.

Many open-source projects still haven't decided on a policy, but have discussed about the topic:

• WIP Linux kernel policy
• NetBSD policy, WIP FreeBSD policy
• Mesa policy (more to come)
• GNU Binutils policy (maybe will be adopted by GCC and glibc)
• Policy for my own projects

[cornelius]

That's a good point. I think an OpenRail policy would be useful as a resource for projects to adopt. We can take that on the agenda of the technical committee and discuss how we would proceed here.

[cornelius]

From issue #299:

AI tools play an increasingly important role in developing software. There are a number of questions how to deal with that in the context of OpenRail open source projects, for example:

• What are the expectations on contributions created with the help or by generative AI tools?
• To what extent in in which way can and should AI be used for reviewing contributions and other code?
• To what extent and under which conditions does it have to be disclosed that AI is used in creating code?
• How to make sure that contributions are properly licensed and contributors have the rights they need to do so?
• How to avoid that projects and individuals get overwhelmed by AI generated code, issues, communication?
• Where lies responsibility for AI generated artifacts?
• What dynamics in the community can be positively influenced with the help of AI? What dynamics might be negatively influenced?
• It could make sense to have an OpenRail wide policy which provides a framework to answering these and other related questions. This issue is meant for collecting some input and thoughts to discuss the need, scope and content of such a policy.

There are examples of similar policies:

• https://www.linuxfoundation.org/legal/generative-ai
• https://openinfra.org/legal/ai-policy
• https://github.com/openrobotics/osrf-policies-and-procedures/blob/main/OSRF%20Policy%20on%20the%20Use%20of%20Generative%20Tools%20(%E2%80%9CGenerative%20AI%E2%80%9D)%20in%20Contributions.md

[cornelius]

More policies: https://github.com/melissawm/open-source-ai-contribution-policies