Zig bans LLM-authored issues, PRs, and comments because the project bets on growing contributors, not landing code.
Key Takeaways
Zig’s policy covers all LLM use in issues, PRs, and bug tracker comments, including AI-assisted translation.
The core argument from ZSF VP Loris Cro: reviewing a PR is an investment in a contributor, not in the code; LLMs break that ROI entirely.
Bun, the most prominent Zig-based project (acquired by Anthropic in Dec 2025), runs its own Zig fork and explicitly will not upstream a 4x compile-speed improvement because of the AI ban.
The “contributor poker” framing: maintainers bet on the person, not the PR contents; LLM-mediated contributions make that bet impossible.
Zig received PRs that claimed no LLM use but did, plus hallucination-filled drive-by PRs and 10,000-line first contributions.
Hacker News Comment Review
Commenters largely endorsed the “why review an LLM PR when maintainers can run the LLM themselves” argument, seeing it as the sharpest practical case against AI contributions.
A significant thread questioned whether Bun’s 4x compile improvement is actually blocked by the AI policy or by the PR’s own architectural complexity introducing unhealthy parallelism complexity into the compiler.
Skeptics warned the policy may exclude an entire generation of developers who grow up using AI assistance, effectively capping Zig’s contributor pool long-term.
Notable Comments
@hitekker: argues the real blocker for Bun’s upstream PR is code quality and unwanted complexity, not just the AI policy.
@nayroclade: warns the policy wins short-term contributor quality but may permanently exclude the next generation of developers.
@lccerina: draws a parallel to ZeroMQ’s collective ownership model, where community health is explicitly prioritized over code throughput.
