Zig bans LLM-authored issues, PRs, and comments because reviewer time builds contributors, not code.
Key Takeaways
Zig Software Foundation VP Loris Cro frames review as “contributor poker”: you bet on the person, not the PR contents.
The ban is strategic, not moral: LLM-assisted PRs consume core team review time without growing trusted, long-term contributors.
Bun, the most prominent Zig-based project (acquired by Anthropic in December 2025), uses heavy AI assistance and maintains its own Zig fork.
Bun recently hit a 4x compile performance gain via parallel semantic analysis and multiple LLVM codegen units, but will not upstream it due to Zig’s LLM ban.
Zig explicitly allows non-English contributions and discourages LLM translation, asking reporters to post in their native language instead.
Hacker News Comment Review
Limited discussion so far; one commenter noted that deerflow.tech takes a similar position, restricting LLM contributions unless the author is already known and trusted by maintainers.
The emerging consensus across projects appears to be trust-gating rather than blanket bans: provenance of the contributor matters more than the content of the diff.
Notable Comments
@jwzxgo: deerflow.tech applies the same policy, “unless it’s coming from a known and trusted developer.”
