Their Reasons
The Zig team's argument doesn't rely on emotion. The first layer is legal: the copyright status of AI-generated code is murky, and for an infrastructure project distributed under a permissive license, mixing in code of unknown provenance is a long-term hazard. The second layer is quality economics: reviewing a plausible-looking AI patch often costs a maintainer more time than writing it themselves, and the bottleneck of contribution has never been output but review. The third layer is the most interesting: the process of submitting a patch is itself the contributor's learning process, and skipping it means the community can't grow its next generation of core maintainers.
A Controlled Experiment
Zig's stance is only interesting placed alongside the other end of the spectrum: in the same period, some old-guard projects generously accept hundreds of AI-involved submissions, while others shut AI out. Neither side is foolish—the difference comes from the project's situation. Zig is language-level infrastructure with extremely low tolerance for errors, and its scarcest resource is people who deeply understand the compiler, not lines of code. The open source world is answering the same question in its own ways, and looking back in a few years to see which strategy raised a healthier community will be a very valuable natural experiment.
via: Hacker News