The Godot Foundation, an open-source game engine team, is cracking down on AI-generated pull requests and text, calling human-to-human respect vital for its volunteer-driven community.
On 1 July, 2026, The Godot Foundation announced that it is rewriting its contributor guidelines to severely restrict and ban almost all use of generative AI in code submissions and pull requests (PRs) to its core repository. The rise of advanced generative AI tools has made it incredibly easy for users to “vibe-code”—generating complex code snippets via prompts without actually understanding how they work.
This has resulted in an overwhelming flood of low-quality, buggy, or fundamentally broken PRs. While AI has dropped the cost and effort of submitting code to zero, the cost of reviewing it remains high. Godot’s limited team of volunteer human maintainers are being completely buried under an unsustainable workload, sifting through low-effort automated code.
Under the new draft rules, anyone classified as a “new contributor” (defined as a user with three or fewer already merged PRs) is blocked from proposing major new features or significant code refactors without obtaining explicit, prior permission from a maintainer. The rules mandate that all human-to-human project discussions, GitHub issues, and design rationales must be written directly by the human contributor.
Submitting PRs via autonomous AI agents or trying to bypass the rules will trigger immediate, unappealable bans from the official Godot GitHub repositories. Minor, menial AI assistance is tolerated only if it does not write the actual logic. This includes single-line text/code completion, regex formatting assistance, mass search-and-replace, and using AI tools strictly to translate the contributor’s own human-written text into English. However, if a contributor uses AI in any of these minor capacities, they must transparently disclose it in their PR description.
Maintainers explicitly noted: “AI cannot take responsibility, and we can’t trust heavy users of AI to understand their code enough to fix it.” On the point of banning AI-generated text in human communications, the Foundation explicitly called it a matter of basic respect, stating that because maintainers donate their free time to look at issues and proposals, they expect to converse with a real human. Furthermore, they found it deeply demoralising to spend their limited spare time providing detailed code feedback, only for that effort to be absorbed by an unlearning machine rather than helping a human developer grow.
