Five Rust Project Teams Ban AI-Generated Code to Combat 'Slop'

Five teams within the Rust project have adopted a specific policy governing the use of Large Language Models (LLMs) for contributions to the `rust-lang/rust` monorepo. The move aims to balance the productivity of AI with the rigorous human oversight required for critical systems programming.

Under the new guidelines, contributors may use LLMs for a variety of supportive tasks, including answering questions, analyzing, distilling, refining, checking, suggesting, and reviewing code. However, the policy explicitly prohibits using LLMs to "create" the primary contribution. The project requires that human authors maintain a deep understanding of any code they submit, ensuring that the logic is vetted by a person rather than generated by a prompt.

The Battle Against AI Slop

This shift comes as open-source maintainers increasingly struggle with "AI slop"—code that appears syntactically correct but is not fully understood by the person submitting it. In a project like Rust, where memory safety and high performance are paramount, the burden of verifying AI-generated code often falls on maintainers, potentially degrading the long-term health of the codebase. By restricting LLMs to an analytical role, the teams intend to prevent the outsourcing of cognitive labor to the review process.

Implications for Infrastructure

The policy reflects a growing tension in the developer community between the speed of AI generation and the necessity of human expertise in critical infrastructure. By banning LLM-led creation while allowing LLM-led analysis, the Rust teams are attempting to ensure that contributors remain "capable pilots." The goal is to keep developers engaged in the actual problem-solving process rather than simply passing off the tedium of coding to reviewers via automated generation.

Scope and Conduct

It is important to note that this is a team-level agreement and does not represent an official, project-wide stance that applies to all areas of the Rust ecosystem. Despite the restrictions on code creation, the policy includes a strict conduct clause: it explicitly forbids harassing contributors for using LLMs, regardless of whether that use is permitted under the current guidelines. Observers will be watching to see if this team-level experiment eventually scales into a broader project standard.