Godot’s AI Ban Is Really About People, Not Machines
An open source AI ban is a project rule that restricts or forbids AI-generated code, text, or autonomous coding agents from contributing to a shared codebase in order to protect reviewer capacity, preserve accountability for defects, and keep contributor mentoring focused on humans rather than tools.
When the Godot Foundation moved to prohibit autonomous coding agents, “vibe-coded” submissions, and any substantial AI-generated code, it was framed as quality control. In reality, it is a blunt response to a maintainer burnout crisis that AI has accelerated. The team is rewriting its GitHub contribution policy to demand that “all code be human authored,” block agent-written comments in discussions, and auto-ban obvious agent spam. New contributors with three or fewer merged pull requests now need explicit approval before landing big features or refactors, a clear attempt to filter out low-effort slop while keeping the door open to people who want to learn the engine. This is not a philosophical rejection of AI. It is a tired group of humans saying: enough.
Demoralizing PRs And The Collapse Of Mentoring
Maintainers are not complaining about AI because it offends some coding purity; they are complaining because it feels like being turned into unpaid tech support for spam. Godot maintainers describe AI pull requests as “demoralizing,” draining wastes of time that pile on top of an already heavy backlog. One studio bluntly called most AI PRs “largely garbage” from users who do not understand what they are proposing. Worse, the people pasting in model output often cannot fix the bugs they introduce, because, as the project put it, “AI cannot take responsibility, and we can’t trust heavy users of AI to understand their code enough to fix it.”
In a healthy project, review is slow but satisfying work because it is also teaching. Reviewers invest hours knowing that today’s newcomer might be tomorrow’s maintainer. AI breaks that bargain. When feedback flows into a black box that cannot grow into a human peer, the moral justification for nights spent on GitHub review queues falls apart. This is why the Godot team wants contribution discussions to stay human-to-human and why they are willing to block AI-generated text in those channels except for translation. The code is less the issue than the loss of a shared learning loop.
What The Data Shows: AI Isn’t Killing Newcomers, It’s Overloading Maintainers
The fear many maintainers voice is that autonomous coding agents will steal the beginner-friendly tickets that used to bring humans into open source. A large study from Peking University tracked 1,888 GitHub repositories that added AI agents like Cursor and Claude Code and compared them with similar projects that did not. Newcomer participation did not collapse; it stayed steady or inched up, with the worst observed decline at 1.5%, too small to be statistically meaningful. In other words, AI is not yet pushing humans out of entry-level contributions. The real change is on the other side of the pull request.
“Pull request volume has nearly quadrupled,” the study reports, while the number of maintainers has not grown at the same pace. Cyclomatic complexity rose 3% to 4% across languages, and Python projects saw cognitive complexity climb around 11% after adoption. The people submitting this more complex code often “may not fully grasp what they’re proposing.” GitHub itself has had to respond with new limits on how many open pull requests outsiders can have at once and tools to triage swollen review queues. Godot’s open source AI ban is not an outlier; it is an early, explicit attempt to regulate a tidal wave others are still trying to redirect.
Torvalds’ Reframe: Tools Don’t Change Responsibility
Linus Torvalds cuts through much of the rhetoric by pointing out that AI-generated source sits in the same abstraction stack as compilers and assemblers. In his words, “AI is a great tool, but it’s a tool.” We already accept that “one hundred percent” of our code is written by compilers without stripping humans of authorship; AI, he argues, is another layer translating human intent into machine instructions. He is “one hundred percent convinced that AI is changing programming, but it’s not changing the fundamentals,” which still include problem decomposition and architectural judgment.
That perspective exposes the real fault line in the Godot debate. Maintainers are not afraid of faster editors. They are afraid of losing any clear line of responsibility in a world where code hops from human to AI to compiler. A tool cannot be shamed into writing tests, nor can it answer for production failures. People can. Torvalds’ focus on human responsibility and people skills is an implicit rebuke to both AI hype and AI panic: the question is not whether we use autonomous coding agents, but who stands behind their output when it reaches the main branch.
The Sustainability Crisis Hiding Behind AI Slop
The uncomfortable truth is that the open source maintainer burnout crisis predates AI; models have simply exposed how fragile the system already was. Godot’s foundation admits that the number of qualified reviewers is small, reviewing pull requests is demanding, and they “can’t keep up with everything coming in.” AI-generated contributions “compound” that by dropping the cost of opening a pull request to nearly zero while leaving review effort unchanged. GitHub-wide, merged pull requests have ballooned from about 25 million per month in early 2023 to roughly 90 million a month today. The math is brutal: far more code, slightly more complexity, same number of volunteers.
Godot’s response is to raise the cost of participation: explicit approval for big changes from newcomers, bans on autonomous agents and vibe coding, and a strict demand that code and discussion remain human-made. The policy is conservative by design, and the foundation says it will revisit it as tools evolve. Critics will call this reactionary. In reality, it is a stopgap. The Peking University researchers say future work must study projects born with AI from day one and measure how dependent they are on these tools. Until we build better ways to fund and staff maintenance, more projects will be tempted to follow Godot’s lead—not because they hate AI, but because they are tired of drowning in its output.
