Why Godot Closes the Door to AI-Generated Code

The open-source game engine Godot published an update to its contribution policy on June 30, 2026, restricting generative AI usage in code submissions. The new rule states that code must be human-written and human-reviewed, with AI allowed only for menial tasks such as code completion, regular expressions, and search-and-replace operations, provided contributors disclose AI use in pull request discussions. The policy explicitly bans autonomous AI agents and 'vibe coding' (leading to automatic ban from the GitHub repository), substantial AI-generated code, and AI-generated text in interpersonal communications. Human review of all pull requests is now mandatory. Maintainers had earlier described the influx of AI-generated contributions as 'exhausting and demoralizing,' citing a low number of qualified reviewers and a rising volume of PRs. Maintainer Rémi Verschelde told Game Developer that only about 47 of roughly 3,700 PRs in the last two release cycles disclosed AI use (about 1.3%), but the cost is disproportionate. Godot is not alone: curl ended its bug bounty program in January 2026 due to AI-generated vulnerability reports, with real reports falling from over 15% to under 5%; tldraw began automatically closing all external contributions in early 2026. A December 2025 study by CodeRabbit analyzing 470 GitHub PRs found AI-assisted contributions have about 1.7 times more issues than human ones (10.83 vs 6.45 average reports per PR). The decision highlights that AI shifts costs: generating code becomes nearly free, but reviewing and maintaining it remains expensive human work. Godot's policy protects the sustainability of its contributor model, emphasizing that in open source, the question is not who wrote the code but who can be responsible for it.