Godot Plans AI Code Contribution Ban As Pull Request Reviews Strain

Godot maintainers are preparing a stricter AI contribution policy after saying the open-source game engine has been hit by pull requests that appear to be generated by people who do not understand the code they submit.

The policy is not yet formally amended, but the project has already set out the direction: heavy AI use in substantial code contributions will be treated as a governance and review problem, not simply as a productivity tool.

New Contributors Need Maintainer Permission

Godot maintainers said new contributors will need explicit permission before sending new features or significant refactoring to the project.

The rule applies to contributors with three or fewer merged pull requests.

The team said the limit is meant to keep AI agents and so-called vibe coders out of the main code review flow while supporting contributors who understand the codebase and can work with maintainers.

The maintainers also said contribution discussions must remain human-to-human, with AI agents and bots excluded except for language translation.

The public record lists the policy a practical test of how open-source projects handle AI-assisted software work.

Godot is not banning every tool-assisted workflow, but it is drawing a line around accountability, code review and whether a submitter can answer questions about the change.

Substantial AI-Written Code Faces A Ban

The Godot team said autonomous agent-written contributions and vibe-coded submissions will continue to lead to automatic bans from the project’s GitHub repository.

The updated policy will also prohibit using AI to generate substantial pieces of code.

The maintainers said limited AI help can cover small editing, completion and search-and-replace tasks.

They described acceptable AI use as minor support, and said contributors must disclose any AI role in writing code during the pull request discussion.

Godot maintainers said code review depends on a human contributor being able to explain, maintain and revise the work.

They said AI cannot take responsibility for code, and they said heavy AI users may not understand enough of what they submit to fix it after review.

Pull Request Review Remains The Bottleneck

The Register reported that Godot maintainers had already described AI pull requests as demoralising earlier in the year.

The proposed policy would limit who can bring larger changes and require disclosure when AI helps produce code.

The article also cited wider criticism of vibe coding after examples involving deleted databases and wiped drives, and it quoted Infosys chairman Nandan Nilekani saying professional software development includes more than writing code.

The public record still lacks the final amended policy text, given an exact release date, named how many AI-generated pull requests were rejected, or specified how maintainers will verify undisclosed AI-written code.