Zig bans – Zig, the open-source programming language maintained by a 501(c)(3) and a contributor network, has banned AI-assisted code submissions. Zig President Andrew Kelley said AI-generated contributions are “invariably garbage,” citing a backlog of open pull requests
Zig’s repository has started drawing a line in code — and Andrew Kelley says he’s doing it because the review queue is already too heavy.
On the JetBrains podcast, the Zig President described AI-assisted submissions as “invariably garbage,” saying “People are sending us contributions that have no value whatsoever.” Kelley added that they carry “negative value” because they “take review time away from the team.”
Zig’s ban is direct. The open-source language — maintained by a 501(c)(3) and a network of contributors — allows programmers to submit code as long as they follow its code of conduct. One rule in that system is an outright cutoff for AI-assisted work. Zig will accept no LLM-generated content, nothing paraphrased from an LLM, and nothing edited, brainstormed, or debugged by an LLM.
For Kelley, the issue isn’t just code quality. It’s the structure of the review process. He said code contributions are reviewed by a handful of core team members. creating what he called a “bottleneck. ” because there are far more pull requests than reviewers. At the time of the recording, Kelley said Zig had 200 open pull requests. In that setting, he argued that AI-generated “slop contributions” slow the whole team down even more. “We’ve wasted everybody’s time,” he said.
The language’s mission also matters to how Kelley frames the decision. Zig is “relatively small,” yet it has had an outsize impact. The language was used to create Bun, which was later acquired by Anthropic. Kelley said mentorship is part of Zig’s core mission — and that means AI contributions can work against it. “We’re all trying to get better at programming,” he said. “People who are sending AI pull requests, those people are not helping this goal.”.
Kelley called those contributors “drive-by contributors,” describing programmers who may submit a pull request or two but “will never join the core team.”
The policy is also designed to be enforceable. Kelley said that if he tried to accept only “good” AI pull requests, reviewers would be forced to judge each one. “If I say none whatsoever, then it’s a very easy policy to enforce,” he said.
Zig’s stance comes at a moment when AI coding tools have become mainstream in technology circles. AI-assisted code has “ripped through Silicon Valley,” fueled by tools like Claude Code and OpenAI’s Codex. Some developers use AI to edit or modify their code; others use it to draft it entirely. Big Tech companies have projected goals for the percentage of code written with AI — and some of that share is already happening.
But Zig’s approach is rooted in a different standard than corporate efficiency targets. The language doesn’t have a mandate to be maximally efficient like large public companies, Kelley said. Instead, it aims to keep contributions tied to learning and participation.
The AI ban has already had ripple effects beyond Zig itself, including drama between Bun and Zig after the policy was announced.
Inside Zig’s workflow. the message is now spelled out in policy: keep AI out of submissions. and keep the review time for contributors who will stay. For a project where a small core team is already managing 200 open pull requests. Kelley’s argument is that accepting AI content would only deepen the delay — and make it harder for the community to build anything lasting.
Zig Andrew Kelley AI code ban open-source 501(c)(3) pull requests JetBrains podcast LLM-generated code Claude Code OpenAI Codex Bun Anthropic