Andrew Kelley, the creator of the Zig programming language, has taken positions on two issues that have divided the open-source community: banning AI-generated code contributions from his project and migrating the entire repository from GitHub to a German nonprofit platform. In an interview with JetBrains, Kelley laid out the reasoning behind both decisions with unusual directness.
Why Zig Rejects AI Contributions
The Zig project maintains a formal, categorical ban on AI-generated code. Kelley identified two core reasons. The first is practical: AI-generated submissions consume review time without producing quality output. He described the pattern as predictable, where automated tools paste instructions back from a chat interface, launder them to appear as human-written code, and submit them as pull requests. After several reviews, the project team recognizes that the contributor has no understanding of the codebase and the work amounts to wasted time for everyone involved.
Kelley framed the second reason as a matter of mission. Zig's code review process functions as a mentorship pipeline. Limited reviewer bandwidth is invested in contributors who are likely to grow into core team members. He described this as "contributor poker," a strategy of betting finite review time on people who will stay and learn versus those who are passing through. AI-assisted contributors fall squarely in the second category, he argued, because they are not developing the engineering skills required to contribute meaningfully in the future.
This position stands in sharp contrast to the approach taken by Bun, another open-source project. Bun recently rewrote its entire codebase from Zig to Rust with extensive support from AI agents that ported code, verified correctness, and proposed fixes. Kelley criticized that decision not on technical grounds but on values, stating that the divergence between the two projects came down to fundamentally different priorities rather than language features.
The Migration from GitHub to Codeberg
The same interview revealed that Zig's repository has moved from GitHub to Codeberg, a German nonprofit platform. Kelley cited persistent failures in GitHub Actions as the immediate trigger. The continuous integration pipeline had become unreliable, and after repeated outages, the team decided the infrastructure was no longer functioning for their needs.
Beyond the technical failures, Kelley expressed a broader preference for nonprofit governance over corporate platforms. He noted that nonprofits tend to prioritize stability over rapid growth and monetization, which aligns better with the needs of open-source projects. He contrasted that with corporations constantly chasing the next quarter's profitability and the next product cycle, calling the nonprofit model more predictable and durable.
Some developers noted that GitHub's declining performance coincided with an exponential increase in AI-driven workloads on the platform, though Kelley did not explicitly draw that connection in his remarks.
Why Zig Exists in the First Place
Kelley recounted how Zig came into being while he was trying to build a native digital audio workstation. Each mainstream language he considered failed for a specific reason. JavaScript lacked low-level hardware access, Go introduced stop-the-world garbage collection pauses that broke real-time audio requirements, Rust's borrow checker created friction that derailed UI font rendering work for weeks, and C++ produced persistent memory corruption bugs that consumed extensive debugging effort.
Zig was designed to solve these specific problems. It rejects automatic memory management in favor of explicit allocators, giving developers direct control over how and when memory is used. Kelley quit his job in 2018 to build the language full-time.
The language has since found adoption in notable projects. Ghostty, a terminal emulator, uses Zig. So does TigerBeetle, a low-latency financial database, and Uber's cross-compilation infrastructure.
The Larger Implications
Kelley's stance places Zig at one extreme of a debate that is playing out across open-source projects. Some teams are embracing AI contributions to accelerate development, while others are rejecting them outright to preserve code quality and cultivate a community of engineers who understand the systems they build. The migration from GitHub to Codeberg adds another dimension, reflecting a growing willingness among open-source maintainers to move infrastructure away from platforms they view as increasingly unreliable or misaligned with their needs.
Whether these positions prove sustainable or isolate the project from broader contributions remains to be seen. What is clear is that Andrew Kelley has made his choices deliberately, and he is willing to defend them publicly.