Rust bars LLM-generated code from rust-lang/rust unless pre-arranged and disclosed
Five Rust teams adopted a policy for rust-lang/rust drawing one line: LLMs can answer, analyze, check, and review, but not create. Contributors must pre-arrange, test, and disclose LLM-generated code, soundness-critical changes need a domain-expert author, and reviewers can now close non-compliant PRs against a 1,281-PR backlog.
Source: blog.rust-lang.org ↗
Polished PRs no longer indicate effort; authors of polished PRs no longer necessarily understand their code.
Rust policy author (blog.rust-lang.org)
Why this matters
- → Formalizes LLM code-contribution rules to prevent review bandwidth collapse.
- → Signals that polished code no longer proves effort or understanding.
- → Closes loophole: 1,281 open PRs + easier LLM writing = unmanageable review load.
Rust's AI firewall