The Zig project's rationale for their firm anti-AI contribution policy

2026-04-30 · Source: Simon Willison's Weblog · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

The Zig programming language project maintains one of the most stringent anti-LLM policies in open-source, explicitly banning LLM use for issues, pull requests, and bug tracker comments, including translations. This policy extends even to projects like Bun, the JavaScript runtime acquired by Anthropic, which uses AI assistance and operates its own Zig fork. Bun recently achieved a 4x performance improvement in its compile process by adding parallel semantic analysis and multiple codegen units to the LLVM backend but does not plan to upstream these changes due to Zig's strict ban. Loris Cro, VP of Community for the Zig Software Foundation, articulates the rationale: Zig prioritizes investing in and growing new human contributors over merely accepting code, viewing each contributor as a long-term investment. LLM-assisted contributions, even if perfect, do not foster this growth, as the review time does not cultivate new, confident, and trustworthy human contributors.

Key takeaway

For CTOs and VP of Engineering evaluating open-source project contributions, Zig's "contributor poker" philosophy suggests a critical re-evaluation of how your team engages with AI-assisted submissions. Prioritize fostering human talent and long-term community growth over the immediate efficiency gains of LLM-generated code. Your investment in human mentorship yields more robust and sustainable project health than merely accepting perfect, but unauthored, pull requests.

Key insights

Zig's strict anti-LLM policy prioritizes human contributor development over immediate code contributions.

Principles

• Invest in contributors, not just contributions.
• Reviewing LLM-generated code doesn't grow human talent.

Topics

• Zig Project
• Anti-LLM Policy
• Open-Source Contributions
• Bun JavaScript Runtime
• Anthropic Acquisition

Code references

• oven-sh/zig

Best for: CTO, VP of Engineering/Data, Software Engineer, Director of AI/ML, Tech Journalist

Related on AIssential

• Oracle's OpenJDK Bans Generative AI Contributions While Oracle's GraalVM Allows Them — Oracle projects demonstrate divergent approaches to generative AI in open-source, highlighting IP and accountability challenges.
• A Lawyer Just Beat 500 Developers at Anthropic’s Hackathon — Domain expertise and problem-solving skills are more critical for AI product success than coding ability.
• Quoting Tim Schilling — Tim Schilling argues that using Large Language Models (LLMs) to contribute to open-source projects like Django without genuine human understanding of tickets, solutions, or feedback is detrimental to the community.
• Accelerate your Career with Open-Source AI — Open source profoundly accelerates careers and fosters innovation, but faces significant sustainability and definitional challenges, especially in the generative AI era.
• Maybe the open-source race is splitting into different kinds of “useful intelligence” now — The LLM landscape is splitting into specialized intelligences optimized for distinct tasks, not a single generalized model.
• Making Dagster Easier to Contribute to in an AI-Driven World — AI tools ease PR generation, but effective open-source contribution requires streamlined review and clear guidelines.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.