AI bans are not about fear of tools—they are about saving people’s time
An AI code generation ban is a project policy that restricts or prohibits substantial code written by AI tools from being contributed to a codebase, usually to protect code quality, reduce review overhead, and preserve human learning and mentoring in open source communities. Godot Engine’s new rules against AI-generated code are not a nostalgic backlash; they are a blunt admission that maintainer time is the scarcest resource in open source, and that machines are burning it up. The Godot Foundation is rewriting its contribution policy to bar most AI-generated code from its repositories, prohibiting AI from generating any substantial piece of code even when humans paste and review it. This move follows months of discussion as maintainers report they can no longer keep pace with a growing backlog of pull requests, many of them AI-authored. In other words, the ban is less about what AI can do and more about what humans can no longer afford to do for free.
Godot’s new contribution rules: triage for a drowning review queue
Godot’s policy is surgical rather than Luddite. Autonomous AI agents and “vibe-coded” pull requests already triggered automatic bans; now that stance is being written directly into the Godot contribution rules. The update bans any substantial AI-generated code, whether it comes from a bot or from a human copy‑pasting AI output, even when the contributor discloses and reviews it. AI is still allowed for narrow tasks like code completion, regex, or find‑and‑replace, but only with disclosure in the pull request. Text generated by AI in maintainer discussions is also off‑limits, aside from machine translation of human-written text. Announced on June 30, the rules are meant to reduce review workloads while building a stronger maintainer pipeline. New contributors—anyone with three or fewer merged pull requests—must now get explicit approval before submitting new features or large refactors and are encouraged to focus on bug fixes and documentation first.
Mentorship, not modules: why AI pull requests feel demoralizing
Godot’s most controversial argument is not about AI-generated code quality; it is about the collapse of mentoring value. Reviewing pull requests has always been tedious, but the Foundation points out that this slog historically doubled as training for future maintainers. Feedback on a human’s patch can turn a first‑time contributor into tomorrow’s reviewer. Feedback on a model’s output goes nowhere. As the Foundation bluntly puts it, “AI contributions have the added pain of being demoralizing,” because maintainers cannot trust heavy AI users to understand their own code well enough to apply feedback. The result is that the informal pipeline that turns a first‑time contributor into a future maintainer stops functioning when the contributor is effectively a machine. Godot frames its crackdown as necessary triage: “We need to take steps to reduce the burden on maintainers while ensuring we still have a pipeline to mentor new contributors to become future maintainers.”
The data: AI isn’t pushing newcomers out, but it is flooding maintainers
A recent study from researchers at Peking University complicates the narrative that AI coding agents automatically destroy newcomer pipelines. Tracking 1,888 repositories that adopted tools like Cursor and Claude Code, they found newcomer participation held steady or even ticked slightly upward, with the worst estimate showing a non‑significant 1.5% dip. On the surface, this undercuts fears that AI is displacing beginners. The catch is maintainer workload. Cyclomatic complexity rose 3% to 4% across languages, while cognitive complexity in Python climbed about 11% after AI adoption. At the same time, pull request volume has nearly quadrupled, and GitHub reports merged pull requests rising from about 25 million per month in early 2023 to roughly 90 million a month today. Contributors may not fully understand what they are proposing, yet every extra branch and nested block lands on maintainers whose ranks are not growing at the same pace.
The trade-off: automation efficiency vs. a community that can sustain itself
The uncomfortable lesson from Godot’s AI code generation ban is that open source now faces a hard trade-off: accept AI-fueled productivity and higher complexity, or protect a slower but sustainable, community-driven development culture. The study on AI agents suggests “the crowding-out fear is, for now, put to rest” and that AI is not yet pushing newcomers out of established projects. But it also underscores that pull request volume has nearly quadrupled, while maintainer capacity has not. Godot’s answer is to prioritize mentoring humans over processing machine‑authored patches, even if that means forbidding some automation. New contributor approval requirements aim to cut review backlog while keeping learning opportunities open for real people, not their tools. The Foundation describes its current approach as conservative and expects to revisit the policy as AI tools change. For now, the clearest signal to new contributors is this: you are welcome; your AI, less so.
