{"slug": "godot-bans-ai-code-what-open-source-learned-the-hard-way", "title": "Godot Bans AI Code: What Open Source Learned the Hard Way", "summary": "Godot Engine formally banned AI-generated code from its open-source project on June 30, citing maintainer burnout and the inability of AI-reliant contributors to fix their own code. The policy prohibits AI-authored code, autonomous AI pull requests, and AI-generated communication, with automatic bans for violations. The move reflects a growing crisis in open source as AI tools flood repositories with low-quality contributions that externalize review costs onto volunteers.", "body_md": "On June 30, Godot Engine — one of the most popular open-source game engines in the world — formalized what its maintainers had been doing informally for months: banning AI-generated code from the project. The [updated contribution policies](https://godotengine.org/article/contribution-policy-2026/) are specific, the enforcement is automatic, and the reasoning from the Godot Foundation is blunt. Open source finally drew a line on the Godot AI code ban — and it sets a template for what comes next.\n\n## What the Policy Actually Says\n\nThe ban is not a vague statement of values. Godot’s new rules prohibit three things explicitly: AI-generated code in any substantial capacity, pull requests submitted by autonomous AI agents, and AI-generated text in human-to-human communication with maintainers. Violating the agent rule triggers an automatic repository ban. No appeal process is mentioned.\n\nWhat’s still permitted is narrow. Code completion for single lines, regex help, debugging lookups, translation assistance — these survive. Lead maintainer Rémi Verschelde put it plainly: the project does “tolerate some AI assistance,” but as he [told Game Developer](https://www.gamedeveloper.com/business/godot-confirms-it-tolerates-some-ai-assistance-but-rejects-vibe-coded-accusations), “‘some’ is load-bearing.” There is no interpretation where this policy is permissive toward AI-generated code.\n\nThere’s an additional rule targeting newcomers specifically: contributors with three or fewer merged PRs cannot submit new features or significant refactors without explicit maintainer approval. The reason, stated directly in [PC Gamer’s coverage](https://www.pcgamer.com/gaming-industry/open-source-game-engine-godot-will-no-longer-accept-ai-authored-code-contributions-we-cant-trust-heavy-users-of-ai-to-understand-their-code-enough-to-fix-it/), is that maintainers “can’t trust heavy users of AI to understand their code enough to fix it.” When a bug surfaces in merged code and the contributor relied on AI to write it, they often can’t diagnose or patch the issue. The contributor disappears; the maintainers inherit the debt.\n\n## The Asymmetry Nobody Fixed\n\nThe deeper issue here is structural. AI coding tools have pushed the cost of writing and submitting a pull request close to zero. However, the cost of reviewing one has not moved. A developer with Cursor or Claude Code running can generate five pull requests in a day with minimal effort. A volunteer maintainer still needs hours to understand, test, challenge, and respond to each one.\n\nGitHub’s own data makes this concrete: merged pull requests across the platform grew from 25 million per month in January 2023 to 90 million per month in March 2026 — a 3.6x increase in three years. Most of that growth isn’t developers becoming three times more productive. It’s AI-assisted submission volume with costs externalized onto whoever reviews the work.\n\nThe Godot Foundation described the effect on their team as “increasingly draining and demoralizing.” Earlier this year, they admitted publicly they didn’t know how long they could keep it up. The PR backlog grew until it became, in their own words, “a meme in the community.” The formal policy is the result of months of informal rejection eventually becoming unsustainable.\n\n## Godot Is Not Alone, But It’s the Most Direct\n\nThe AI slop problem in open source predates this announcement. Earlier this year, [GitHub shipped a PR throttle tool](https://www.coderabbit.ai/blog/github-gives-maintainers-a-throttle-for-the-ai-pull-request) — exactly two weeks before Godot’s ban — letting maintainers cap how many open PRs a user without write access can submit at once. It addresses volume, not quality. ByteIota covered [curl’s related crisis](https://byteiota.com/curl-takes-july-off-after-ai-slop-killed-its-bug-bounty/) earlier this year, when Daniel Stenberg shut down the bug bounty program after AI-generated fake vulnerability reports dropped the legitimate submission rate below five percent.\n\nWhat Godot did differently is write the policy down. Not as a community norm. Not as a maintainer preference. As a formal rule with automatic consequences. That distinction matters because it creates a template other projects can follow without needing to relitigate the debate from scratch.\n\n## The AI Tool Vendors Didn’t Build for This\n\nThe tools most responsible for the flood — Cursor, GitHub Copilot, Claude Code — were built to measure individual developer productivity. Lines generated per hour. PRs opened. Tasks completed. None of them built friction into the submission path. None of them surface whether a target project accepts AI-generated contributions before you hit submit. And none of them make visible the cost they’re exporting onto volunteer maintainers who didn’t sign up to review machine output.\n\nThat’s not an accident. It’s an optimization choice. Moreover, it’s a choice that open-source projects are now responding to unilaterally, because the tool vendors haven’t. Godot’s ban is a market failure response — when vendors don’t internalize externalities, the downstream communities start building walls.\n\n## What Developers Should Watch\n\nGodot’s policy will not stay unique for long. The reasoning — AI submissions undermine mentorship loops, create legal exposure from possible copyright contamination in training data, and flood volunteer review queues — applies equally to Linux, Apache, and dozens of major open-source projects. Expect formalized policies, not just community norms, from more projects before the year ends.\n\nThe ironic solution gaining traction in some communities: AI-powered code review to detect and auto-reject AI-generated contributions. Fighting machine output with machine detection. Whether that scales, or simply adds another fragile trust layer to an already strained system, remains an open question.\n\nFor developers, the immediate takeaway is clear: check contribution policies before submitting AI-assisted code. For projects, Godot just demonstrated that a formal line in the sand is more effective than hoping the problem resolves itself. It doesn’t.", "url": "https://wpnews.pro/news/godot-bans-ai-code-what-open-source-learned-the-hard-way", "canonical_source": "https://byteiota.com/godot-bans-ai-code-what-open-source-learned-the-hard-way/", "published_at": "2026-07-01 10:10:09+00:00", "updated_at": "2026-07-01 10:31:04.364752+00:00", "lang": "en", "topics": ["ai-policy", "ai-ethics", "developer-tools"], "entities": ["Godot Engine", "Godot Foundation", "Rémi Verschelde", "GitHub", "Cursor", "Claude Code", "Daniel Stenberg", "curl"], "alternates": {"html": "https://wpnews.pro/news/godot-bans-ai-code-what-open-source-learned-the-hard-way", "markdown": "https://wpnews.pro/news/godot-bans-ai-code-what-open-source-learned-the-hard-way.md", "text": "https://wpnews.pro/news/godot-bans-ai-code-what-open-source-learned-the-hard-way.txt", "jsonld": "https://wpnews.pro/news/godot-bans-ai-code-what-open-source-learned-the-hard-way.jsonld"}}