Zig Challenges AI: A Controversial Anti-LLM Policy

⚡

Key Takeaways

1Zig imposes a strict ban on LLMs for contributions, including issues and pull requests.

2Bun, a major project in Zig, has improved its performance but cannot integrate its advancements due to this policy.

3Loris Cro, from the Zig Software Foundation, emphasizes that the goal is to train reliable contributors, not just to accept code.

💡Why it matters — This policy could influence how open source projects manage the integration of AI technologies.

Zig and Its Anti-LLM Policy

The Zig project stands out for having one of the strictest policies against the use of language models (LLMs) among major open-source projects. This ban covers several aspects: LLMs cannot be used to solve problems, write pull requests, or comment on bug tracking. While English is encouraged, contributors are free to express themselves in their native language, leaving it to others to use their own translation tools to understand their contributions.

Bun and Its Fork of Zig

The most prominent project using Zig is undoubtedly the JavaScript engine Bun, which was acquired by Anthropic in December 2025. Bun uses its own fork of Zig and has recently achieved a significant performance improvement, quadrupling compilation speed after integrating parallel semantic analysis and multiple code generation units into the LLVM backend. However, despite these advancements, Bun does not plan to merge these improvements back into Zig due to the strict ban on contributions written by LLMs.

Loris Cro's Article and Zig's Philosophy

In the article titled "Contributor Poker and Zig's AI Ban" published via Lobste.rs, Loris Cro, Vice President of the Zig Software Foundation community, explains the reasons behind this strict ban. He emphasizes that in successful open-source projects, there comes a time when the number of pull requests exceeds the team's capacity to process them. Rather than rejecting imperfect requests, Zig chooses to support new contributors so they can progress, which is not only fair but also strategic.

Investing in Contributors

Zig values contributors more than their contributions. Each contributor represents an investment for the core Zig team. The primary goal of reviewing and accepting pull requests is not simply to add new code, but to train new contributors who can become trustworthy and prolific over time. According to Cro, LLM assistance disrupts this process, as it does not contribute to the integration of new trustworthy contributors.

The Concept of "Contributor Poker"

Loris Cro uses the analogy of poker to describe this approach, stating that, just like in the card game, "you play the person, not the cards." In "contributor poker," the focus is on the contributor rather than the content of their initial pull request. This philosophy raises a crucial question: why would a project maintainer spend time evaluating a request written by an LLM when they could use an LLM to solve the problem themselves?