LLVM contributor Nikita Popov opened an RFC to extend the project’s comparatively permissive AI tool policy by banning AI-generated or substantially AI-assisted textual communication, including pull request descriptions and RFC texts, while leaving the allowed uses of AI for code contributions unchanged. He argues the main cost is the volume of verbose, over-formatted pull request descriptions arriving from new contributors, which consume reviewer bandwidth twice over: once to read and again to rewrite into a mergeable form, and which erase the effort signal that used to help maintainers judge whether a stranger’s contribution was made in good faith. The thread is the latest sign that large projects are separating AI-assisted code, which many still accept with human review, from AI-authored prose, which they increasingly reject.
LLVM proposes barring AI-generated pull request and RFC write-ups
LLVM contributor Nikita Popov opened an RFC to extend the project's comparatively permissive AI tool policy by banning AI-generated or substantially AI-assisted textual communication, including pull request descriptions and RFC texts, while leaving the allowed uses of AI for code contributions unchanged. He argues the main cost is the volume of verbose, over-formatted pull request descriptions arriving from new contributors, which consume reviewer bandwidth twice over: once to read and again to rewrite into a mergeable form, and which erase the effort signal that used to help maintainers judge whether a stranger's contribution was made in good faith. The thread is the latest sign that large projects are separating AI-assisted code, which many still accept with human review, from AI-authored prose, which they increasingly reject.
Source: Llvm