Rust-lang/rust is adopting an LLM policy

A new set of guidelines has been added to the Rust core repository that spells out acceptable practices for using large language models when contributing code or discussion. The document targets four specific roles: PR reviewers, authors who include LLM-generated code, people who surface bugs via LLMs, and commenters who directly quote an LLM. By publishing the rules, the project moves from ad-hoc handling to a transparent standard that newcomers can read before submitting work.

The policy was motivated by three observed problems: polished pull requests no longer reliably signal author effort, the ease of generating code inflates the already high review backlog, and mechanical copy-pasting of LLM output wastes reviewer time. These issues have made it harder for maintainers to assess whether a contribution reflects genuine understanding or merely a quick generation. The guidelines aim to restore the link between code quality and author intent, which is central to Rust's collaborative culture.

For contributors, the immediate cost is the need to annotate any LLM involvement and to be prepared to explain the reasoning behind generated code. This adds a step to the contribution workflow, but it also forces authors to internalize the changes they propose, reducing the likelihood of superficial submissions. Failure to follow the new disclosure requirements can result in a PR being closed without further discussion.

Reviewers gain a concrete reference point that can be cited when rejecting or requesting changes to LLM-heavy submissions. This should reduce ambiguity and the mental load of deciding whether a polished PR truly reflects effort. However, reviewers still need to evaluate the underlying design decisions, as the policy does not replace the substantive judgment required for code acceptance.

The scope of the policy is limited to activities within the rust-lang/rust monorepo and does not extend to other Rust projects or external tooling. It is explicitly described as a non-official stance on LLMs for the broader Rust community, meaning that teams outside the listed groups can continue their existing practices. Consequently, the rule set stops being applicable once a participant is not directly involved in PR review, authoring, issue discovery, or commenting that includes LLM text.