Zig's Anti-AI Policy: A Strategic Bet on Human Contributors

The Human Investment Thesis

The Zig programming language has drawn a definitive line in the sand. Its code of conduct contains one of the most stringent policies in major open source: an explicit ban on using Large Language Models for issues, pull requests, or comments. This stance isn't merely a philosophical objection; it's a calculated strategy rooted in the economics of project maintenance.

As explained by Zig Software Foundation VP Loris Cro, successful projects inevitably face a bottleneck: more pull requests than maintainers can process. The intuitive response might be to only accept perfect contributions. Zig takes the opposite, more labor-intensive path. They invest significant time helping new contributors refine their work, viewing each person as a long-term asset.

This philosophy is termed "contributor poker." The core tenet, as Cro articulates, is that "you bet on the contributor, not on the contents of their first PR." The primary goal of reviewing a pull request isn't to land code; it's to grow a new, trusted collaborator. An LLM-generated PR, no matter how flawless, represents a dead-end investment of review time.

It fails to cultivate the human expertise and community trust that scales a project sustainably. This rationale provides a clear answer to a common question: if a PR is AI-written, why should a maintainer review it instead of using their own AI to solve the problem?

Real-World Impact: The Bun Fork

The practical consequences of this policy are illustrated by Bun, the high-performance JavaScript runtime written in Zig and acquired by Anthropic in late 2025. Bun's development heavily utilizes AI assistance, creating a fundamental conflict with Zig's upstream rules.

Recently, the Bun team achieved a significant 4x performance improvement for Bun's compile times by adding parallel semantic analysis and multiple codegen units to Zig's LLVM backend. Despite this technically valuable contribution, the Bun team stated they "do not currently plan to upstream this, as Zig has a strict ban on LLM-authored contributions."

This situation forces Bun to maintain its own fork of Zig, creating a potential long-term divergence. It highlights a growing schism between AI-native development practices and projects prioritizing human-centric collaboration models. The trade-off is clear: forfeit the immediate efficiency gains of AI to preserve a specific community-building methodology.

A Cultural Movement Beyond Code

Resistance to AI's encroachment extends far beyond programming languages. In the DIY publishing world, zine creators are mounting a vocal defense of handmade art. Illustrators like Melbourne's Maddie Marshall and Philadelphia's Rachel Goldfinger have created dedicated anti-AI zines, such as "I Should Be Allowed To Think," to protest the technology's pressure on creative jobs.

"AI is eliminating a lot of people’s ability to think critically for themselves," says Goldfinger. For these artists, the handcrafted, scrappy essence of zines is fundamentally incompatible with AI automation. Ione Gamble, founder of Polyester zine, now runs all submitted articles through an AI checker to ensure authenticity.

This sentiment echoes in the corporate world. Over 600 Google and DeepMind employees, including principals and VPs, recently signed a letter to CEO Sundar Pichai. They demanded Google refuse Pentagon contracts for classified AI work, citing ethical concerns about lethal autonomous weapons and mass surveillance.

The signees, aware of AI's power to centralize authority, argued their proximity to the technology creates a responsibility to prevent its most dangerous uses. Their protest followed the Pentagon's pressure on Anthropic to ignore its own "red lines" for ethical AI deployment.

The Political Unification of Anti-AI Sentiment

Opposition to AI is crystallizing into a cross-political force. Figures as ideologically disparate as Steve Bannon and Bernie Sanders have found common ground in criticizing the technology's lack of transparency, accountability, and its perceived threat to the working class.

As reported, a growing coalition worries that tech companies are prioritizing profit over societal impact, funneling wealth to Silicon Valley while the middle class bears the costs. This animus stems from a feeling of losing personal autonomy to opaque computer systems.

John Oliver's commentary on AI has further amplified these concerns to a mainstream audience. The core fear is a loss of control and being forced to pay for the disruption AI introduces to lives and livelihoods. This brewing discontent is positioned to influence the political landscape, potentially affecting upcoming elections.

Why This Strategic Stance Matters

Zig's "contributor poker" model is a high-stakes gamble. In an industry racing to adopt AI for productivity, Zig deliberately sacrifices short-term velocity. The bet is that a tightly-knit, highly-skilled community of human experts will produce more robust, innovative, and sustainable software in the long run.

This approach rejects the transactional nature of AI-assisted contributions. It argues that the value of a pull request isn't just the code diff, but the strengthened relationship and increased project capacity it represents. When an LLM is the author, that relational value is zero.

The policy also acts as a cultural filter. It attracts contributors who value deep understanding and craftsmanship over rapid output. This aligns Zig with broader movements in art and ethics where human agency, critical thought, and intentionality are being defended against automated homogenization.

As AI becomes ubiquitous, Zig's choice represents a conscious alternative. It's a case study in whether cultivating human capital, with all its inefficiencies, can compete with the raw output of machine-assisted development. The success of Zig, and the fate of its fork with Bun, will be a critical data point in that ongoing experiment.