The foundation behind the open-source Godot game engine has announced a significant policy shift regarding the use of generative AI by contributors. In response to an influx of "demoralizing" and "low-effort slop" that has exacerbated an already substantial backlog of code review requests, the organization will now require contributors to disclose any use of AI in their submissions.
While AI tools will still be permitted for minor, "menial" tasks such as code completion, finding and replacing text, or generating regular expressions, the foundation is taking a firm stance against "vibe coding" – the practice of using AI to generate large portions of code. This stricter approach aims to maintain the quality of contributions and ensure that the engine's development remains a human-driven endeavor.
Godot is a popular engine used in titles such as Slay the Spire 2, Buckshot Roulette, The Case of the Golden Idol, and Cruelty Squad. Its popularity surged, particularly after rival engine Unity faced backlash for proposing per-install fees. The engine relies heavily on volunteer contributors and seasoned "maintainers" for its ongoing development. Code contributions are submitted as "pull requests" for review, and the core team has acknowledged a growing number of these requests, partly due to the increasing use of generative AI by contributors.
The foundation noted that while the ease of AI-generated code has increased the volume of submissions, the capacity for review has not kept pace. This has created a "reviewer shortage" that can no longer be ignored. Furthermore, the use of AI-generated code presents a challenge to the engine's community-based training model, where experienced volunteers mentor newer contributors. Reviewing AI-generated code offers no such reciprocal benefit, making the already tedious task of code review less rewarding and harder to justify dedicating free time to.
AI models, the foundation explained, cannot learn from specific feedback in the way a human can, nor can they take responsibility for the code they generate. This raises concerns about accountability and the ability to fix issues when they arise. Consequently, the new policies are designed to foster the development of human contributors and ensure that reviewers' time is spent effectively.
The updated rules stipulate several key points: Autonomous AI agent use or "vibe coding" will result in an automatic ban from the GitHub repository. AI should not be used to generate substantial code; human authorship is required, with AI assistance limited to menial tasks. Any use of AI in code authoring must be disclosed in the pull request. AI-generated text should not be used in human-to-human communication with maintainers, as this is seen as a basic principle of respect, though machine translations of human-written content are still acceptable. All pull requests must be reviewed and approved by a human before merging, a practice the foundation is making more explicit in its policy.
In addition to AI-related rules, Godot will now block new features or significant refactoring from new contributors – defined as those with three or fewer approved pull requests – without explicit permission from maintainers. This aims to ensure new contributors learn the codebase and engage with maintainers by working on bug fixes and documentation before undertaking larger projects.
This cautious approach to generative AI by the Godot Foundation stands in contrast to the direction taken by Epic Games with Unreal Engine 6, which is slated to offer extensive integration of AI models. Godot's stance highlights a growing tension within the game development community as it navigates the capabilities and ethical implications of generative AI in creative and technical workflows.