According to The How-To Geek, the Zig programming language, led by foundation president Andrew Kelly, is officially quitting GitHub and moving its main repository to Codeberg. The decision comes after what Kelly describes as a seven-year decline in engineering quality since Microsoft’s acquisition, culminating in a breaking point with the bug-ridden GitHub Actions system. A critical bug in a ‘safe_sleep.sh’ script caused processes to run forever at 100% CPU, taking down CI services for weeks. The team also cites GitHub CEO Thomas Dohmke’s directive to “embrace AI or get out” and the platform’s aggressive push of Copilot, which violates Zig’s strict no-AI policy. While GitHub Sponsors still provides a large chunk of Zig’s revenue, the GitHub repository is now read-only, and the new canonical home is on the non-profit Codeberg platform, with issue numbering starting fresh at 30,000.
The stakes for developers
This isn’t just about one project throwing a fit. It’s a signal flare. For developers, CI/CD pipelines are the circulatory system of modern software. When GitHub Actions starts what Zig calls “vibe-scheduling” jobs at random and basic scripts fail catastrophically, trust evaporates. You’re not just fighting your own code anymore; you’re fighting the platform that’s supposed to help you ship it. And look, every platform has bugs. But the feeling Kelly describes—that priorities have “completely rotted” in favor of bloated JavaScript and AI features—that’s a vibe a lot of devs are sensing. It’s the classic enshittification playbook: attract users with a great core product, then let it decay while you monetize a shiny new thing. In this case, the shiny new thing is Copilot, which Microsoft CEO Satya Nadella says drives 40% of GitHub’s annual revenue growth.
The AI divide is real
Here’s the thing: Zig’s announcement makes their no-AI policy a central pillar of their identity. They see LLM-generated code as a policy violation, and GitHub is shoving Copilot in their face. This creates a fundamental cultural mismatch. For every developer thrilled with AI autocomplete, there’s another who sees it as a vector for security flaws, licensing issues, and a degradation of craft. GitHub betting the farm on AI necessarily means deprioritizing the needs of those in the latter camp. Zig’s exit is the most direct consequence of that bet I’ve seen. And they’re not alone; the Dillo browser project is planning a similar departure for the same reasons. This could be the start of a quiet sorting, where projects with strict code integrity requirements seek out quieter, more focused homes.
The bigger picture for enterprises
So what does this mean for the big guys? For enterprises and large open-source projects, vendor lock-in is a silent tax. GitHub Sponsors is a powerful retention tool, as Zig admits. Leaving means potentially leaving money on the table and fracturing community discussion. But Zig’s migration strategy—leaving old issues behind and starting fresh—shows a brutal pragmatism. It prioritizes a clean future over a messy past. This move should make any large project’s maintainers think: what’s our exit strategy? Is our project’s infrastructure resilient, or are we one corporate policy shift away from crisis? For companies that rely on a stack of open-source tools, this is a reminder that the foundations can move. When core infrastructure languages like Zig make a shift, the ripple effects eventually reach everyone. It’s a stability question.
Is this the beginning?
Will there be a mass exodus? Probably not tomorrow. The network effect and convenience of GitHub are massive. But Zig has provided a blueprint and a powerful justification. They’ve framed it not as a petty complaint, but as a necessary engineering decision to protect their project’s quality and principles. That’s a compelling narrative for other maintainers feeling the same pain. Microsoft and GitHub are clearly banking that the allure of AI and the convenience of an all-in-one platform will outweigh these defections. But they’re sacrificing the trust of the most meticulous builders. And in technology, those are often the people whose work ends up running the world. Betting against them has historically been a bad idea. This feels like a crack in the dam. Let’s see if more follow.
