Five teams in the Rust project have adopted a new policy governing how Large Language Models (LLMs) can be used when contributing to the rust-lang/rust monorepo. The policy, authored by a core team member, is not an official stance on LLMs and applies only to specific contributor groups: those who review or moderate PRs, those who author LLM-generated code PRs, those who report LLM-discovered issues, and those who quote LLMs in issues or comments.

The policy was created due to growing concerns around LLM usage in the community. Previously, LLM use led to three main issues: polished technical products no longer indicating effort or understanding; increased review burden due to easier code generation; and mechanical copy-pasting from LLMs wasting reviewer time. With 1,281 open PRs at the time of writing, the influx of LLM-assisted contributions exacerbated existing review bandwidth problems.

Before the policy, moderation was inconsistent under a "wild west" approach with no disclosure rules. The new policy distinguishes between acceptable LLM uses (answering questions, analyzing, refining, suggesting, reviewing) versus restricted ones (creating content). Uses in the first category require disclosure, while creation-based uses are heavily restricted. The policy aims to make community expectations transparent for new contributors and provide clear guidance for reviewers.

The policy acknowledges internal disagreement about AI use but emphasizes shared values of building deep expertise and inclusivity. It includes provisions for future changes and potential sub-team creation to streamline governance, reducing the need for large approval processes.