You’ve probably spent hours debugging a merge request, only to have someone tell you your code is “illegitimate.” If you’ve ever used AI to help you write software, you’re about to feel that sting. Sourcehut, a beloved open-source hosting platform, just proposed a ban on all “vibe coded” projects — code created with substantial AI assistance. And the reason they’re giving? Quality. Provenance. Copyright. But let’s be honest: this isn’t about protecting free software. It’s about who gets to decide what counts as real development.
The proposal reads like a manifesto. It argues that large language models are “harmful to the free software ecosystem as a whole,” that they appropriate code without respect for licenses or attribution. On the surface, that sounds noble. But scratch just a little, and you’ll see the real story: a small group of gatekeepers trying to draw a line between “authentic” human work and “AI pollution.”
I’ve been in open source for over a decade. I’ve seen forks, flame wars, and license disputes. But this is different. This is a power grab dressed in moral clothing. The moment you tell someone their contribution doesn’t count because it was assisted by a machine, you’re not defending quality — you’re defending your own authority.
One commenter on the proposal put it perfectly: “We are seeing the slow death of social coding.” Another said, “Protected by go-away — so I went away.” That’s the reaction from people who actually use AI to build. They feel excluded. They feel judged. And they’re right to feel that way.
Here’s the uncomfortable truth that Sourcehut’s leadership doesn’t want to say out loud: they’re afraid. Afraid that AI-generated code will flood their repos, making it harder to trace origins, harder to enforce licenses, harder to maintain the curated feel of the platform. But the solution they’re proposing — a blanket ban — is lazy, arrogant, and ultimately destructive.
Because here’s what happens when you ban vibe coding: you push the most productive, curious, and experimental developers elsewhere. You create a platform that only admits code written by people who follow the old rules. You turn open source into a gated community where the password is “I wrote every line myself.” And that’s not open source at all. Inclusivity isn’t about letting everyone in — it’s about not having a gate in the first place.
I get it. AI-generated code can be messy. It can miss attribution. It can produce garbage. But so can human developers. We don’t ban beginners because they write bad code. We mentor them. We review their pull requests. We teach them. Why should AI-assisted code be treated differently?
The real issue isn’t quality. It’s ownership. The people who control the repositories want to keep control. They want to decide what counts, who gets credit, and how the narrative of open source is written. This ban isn’t about protecting the commons — it’s about protecting the curators.
If you’re a developer who uses AI tools — even just for boilerplate, or for brainstorming, or for debugging — you should be paying attention. This proposal is a test. If it passes, other platforms will follow. And soon, the line between “real” developer and “vibe coder” will become a caste system. The haves and the have-nots. The pure and the tainted.
I’m not saying we should ignore copyright or provenance. Of course we need to respect licenses. But we can solve those problems with better tooling, better attribution, better education. We don’t solve them by banning people. Banning is the easy way out. The hard way — the way that actually honors the spirit of open source — is to figure out how to include everyone, including the machines, while keeping the values intact.
Sourcehut has a choice. So do we. We can let the gatekeepers win, and watch open source become a museum of human-only code. Or we can demand a future that’s messy, inclusive, and alive. I know which one I’m choosing. The question is: will you?
FAQ
Q: Isn't banning AI-generated code a reasonable way to protect open source from copyright problems?
A: No, because it's a blunt instrument that punishes all AI-assisted work, even when the developer has verified provenance. Better solutions exist: automated attribution scanning, clear labeling of AI-generated code, and community norms. A ban just excludes people without solving the underlying issues.
Q: What's the practical implication for me as a developer who uses AI tools?
A: If this policy spreads, you'll be locked out of certain platforms unless you lie about how you wrote your code. More importantly, it signals that your work is considered less legitimate. You'll have to choose between using AI and being part of the open source community — a false choice that hurts innovation.
Q: Isn't it fair to say AI-generated code is low quality and should be kept out?
A: That's a stereotype. AI can generate high-quality code, but it's also capable of producing garbage — just like humans. The real issue is not quality but control. The ban is a way for established developers to maintain their authority over what 'real' code looks like. It's gatekeeping, not quality assurance.