Godot bans – The Godot game engine has updated its contribution policy to curb AI-generated pull requests, citing how low-quality PRs pile up faster than volunteer reviewers can handle them. The new rules ban autonomous AI agents, restrict substantial AI code generation, p
The pull requests kept coming. But the code quality didn’t keep pace.
For Godot. the open source game engine built and maintained largely by volunteers. the flood has grown worse as AI-generated code becomes easier to produce. The project says the problem isn’t just that there are more submissions—it’s the inverse relationship between PR code quality and the number of PRs. In practice, that means reviewers are forced to spend limited time sorting through contributions that don’t move discussions forward.
In the announcement. the project also points to a quieter kind of damage: the demoralizing effect that AI-generated pull requests can have on the people doing the reviewing. The human behind a PR, the project writes, may not be interested in being educated. In some cases. it may even be an automated agent that can’t sustain a useful conversation about the tradeoffs in a coding approach—let alone become a consistent maintainer over time.
That mix—volume, mismatch, and stalled discussion—has now pushed Godot to tighten how it accepts contributions.
The update includes several explicit bans. Autonomous AI agents and “vibe coding” are no longer welcome. The policy also bans substantial AI generating of code. And it bars AI-generated text in human-to-human communication, aiming to keep discussion grounded in real participation rather than auto-generated messages. Beyond those restrictions. the project codifies one firm requirement: every pull request must be reviewed and approved by a human being before it can be merged.
Godot’s stance isn’t happening in a vacuum. The new policy is described as similar to what the Mesa project requires. where submitters must be able to comprehend the code they’re contributing. It stops short of NetBSD, which treats LLM-generated code as “tainted” over potential licensing and other concerns. Other communities. including the Linux kernel. have taken a different approach—making the submitter responsible for declaring when an AI tool is used.
Taken together, the pattern is hard to ignore: across open source, tools that once promised to speed up contributions now appear to be reducing useful interactions. Godot’s rules make that tradeoff visible in policy form, not just in hallway conversations.
Whether this is a temporary pivot or the beginning of a broader retrenchment remains to be seen. What is clear is that, for open source projects like Godot, AI-assisted contributions are being pushed toward stricter limits—and increasingly, out of the default path to merge.
Godot contribution policy pull requests AI-generated code open source cybersecurity code review volunteer maintainers Mesa project NetBSD Linux kernel AI tools
So they’re banning AI now? Seems kinda backwards but I guess if the PRs are junk.
I don’t even know what a “vibe coding” is but yeah, people been spamming GitHub with garbage for forever. Still, making it so every PR has a human approval sounds like bureaucracy.
This is gonna end up hurting normal contributors though. Like I’m pretty sure some AI stuff is fine, it’s not like it can steal your code or whatever. Also “banned text in human-to-human communication”?? Does that mean like comments can’t be written by AI even if it’s helpful?
Finally. Those AI PRs always feel like they don’t even understand what they’re changing. But I wonder if this is just because volunteers are overwhelmed and they need a scapegoat. Like the part about demoralizing reviewers… yeah, that’s real. I just don’t get why they can’t just require a real explanation instead of banning the tools completely.