Godot Officially Bans AI-Generated Code and Vibe Coding From All Future Contributions

The Godot Foundation Is Drawing a Hard Line on AI Slop

The Godot Foundation has had enough. On June 30, 2026, the team behind the popular open-source game engine published a new contribution policy that bans AI-generated code, autonomous AI agent contributions, and AI-authored text in maintainer communications. No vibe coding. No pasting ChatGPT output into a pull request. No letting an agent submit PRs on your behalf. If you cannot write the code yourself and take responsibility for it when something breaks, Godot does not want it.

Four Months of Warning Before the Ban

This is not a surprise move. Back in February 2026, Godot co-founder Remi Verschelde went public with a grim situation: the engine's GitHub repository was getting swamped with AI-generated pull requests containing nonsensical code changes, fabricated test results, and the kind of verbose wall-of-text descriptions that are unmistakably LLM output. He said maintainers did not know how long they could keep up with reviewing it all. That post blew up, with the PC Gamer tweet covering it hitting over 13 million likes. The community was clearly watching.

The Foundation spent the following months debating how to respond. The result is a formal policy update that goes beyond just discouraging low-effort submissions. It codifies exactly what will get a contributor auto-banned from the repository. Autonomous AI agents submitting PRs already triggered automatic bans before this policy. That behavior is now spelled out explicitly alongside the broader AI code prohibition.

What the New Policy Actually Says

The updated rules cover three main areas. First, no AI-generated code in any substantial capacity. Contributors can still use AI for small things like code completion, regex patterns, or find-and-replace tasks, but must disclose even that limited use in the pull request discussion. The moment AI is writing meaningful chunks of logic, the PR will be rejected. Second, no AI-generated text when communicating with maintainers. If a reviewer volunteers their weekend to look at your PR, they expect to talk to a person, not get a wall of GPT prose. Third, new contributors with three or fewer merged pull requests will be blocked from submitting major new features or significant refactors without explicit maintainer approval. They need to earn trust by fixing bugs and writing documentation first.

Machine translations are still acceptable, as long as the original text was written by a human. That is a reasonable carve-out. The Foundation is not trying to shut out non-English speakers; it is trying to shut out the flood of submissions where nobody human actually understands what they submitted.

Why Maintainers Were Burning Out

The Foundation's blog post puts the real problem plainly. Reviewing pull requests is already tedious and time-consuming work. The only thing that makes it worth doing is the knowledge that you are helping a new developer learn the codebase and potentially turning them into a future maintainer. When the contributor is an LLM or a person who just pasted AI output without reading it, that mentorship loop collapses entirely. Feedback disappears into a machine. Nobody grows. The reviewer burned an hour of their Saturday on nothing. Do that enough times and maintainers stop reviewing. The backlog grows. The project slows down.

The Foundation made clear that the number of open PRs had already become a running joke in the community before AI made it worse. Godot is used to build real games. Slay the Spire 2 and The Case of the Golden Idol both run on it. The engine needs active maintainers to keep improving, and active maintainers need to feel like their time is going somewhere. AI submissions were actively destroying that feeling.

Community Reaction: Mostly Supportive, Some Skeptics

The reaction online was loud and mostly positive. The PC Gamer tweet covering the ban had 2 million likes and 87 million views within hours. The dev community, which had been watching Godot struggle with AI slop since February, seemed relieved that someone drew a clear line. The concern has never been that AI tools exist; it is that vibe coding encourages people to contribute code they do not understand, creating maintenance debt that someone else has to pay down later.

There are reasonable objections. One developer pointed out that Godot already had trouble keeping up with PR reviews before AI entered the picture, and that adding more barriers to contribution does not solve the reviewer shortage. That is a fair point. The Foundation acknowledges it: the goal is not to reduce the total number of contributions but to shift the mix toward contributions that come from people who understand and own what they submitted. Whether that works in practice depends on how enforceable the ban actually is, since detecting AI-authored code is still far from a solved problem.

A Signal That the Industry Is Paying Attention

Godot is not the first open-source project to push back on AI contributions. But it carries extra weight coming from a game engine with real adoption in the indie space. This is a tool that thousands of developers rely on to ship games. When the team maintaining it says AI slop is degrading the project to the point where maintainers are burning out, that is a concrete consequence, not just ideology.

The broader pattern here is worth watching. AI tools have lowered the cost of submitting code to near zero, but the cost of reviewing that code has stayed the same or gone up. That imbalance does not fix itself. Godot is betting that the right answer is restoring accountability: if you submit code, you wrote it, you understand it, and you can fix it when it breaks. Every other approach just shifts the burden somewhere else.

The CriticalPixel Take

This policy is the right call, and it should have come sooner. The problem was visible in February and everyone knew it was going to get worse. Waiting four months to formalize something that common sense demanded immediately let the burnout compound. The real test is enforcement: detecting AI code without false positives is hard, and determined bad actors will find ways around disclosure requirements. But setting a clear standard matters even before enforcement is perfect, because it gives maintainers a policy to point to when they reject a suspicious PR.

If you are an indie developer who uses Godot and wants to contribute back to the engine, this policy is not hostile to you. It is asking you to do what good contributors were already doing: understand what you submit and stand behind it. The people this policy is actually aimed at are the ones who were treating an open-source project maintained by volunteers as a dumping ground for automated output. Good riddance.