Open Source Says No: 22 Projects Publicly Ban AI-Generated Contributions
If you submit AI-generated PRs to many popular projects, you now risk instant closure and bans, not just review delays. Checking the project's AI policy before contributing is a practical requirement, and the cost of an ignored or rejected PR has gone up.
Open Source Says No (osssaysno.com) lists open-source projects that explicitly ban AI-generated contributions. As of November 11, it verifies 22 projects, each with policy quotes, scope, and a verification date. The site reached the Hacker News front page with 61 points and 106 comments, turning scattered CONTRIBUTING.md notices into a visible ecosystem signal: open source has moved from tolerating AI-generated content to systematically rejecting it.
The bans cover code, documentation, issues, translations, and art. yt-dlp's contribution guide says "NO AI / NO LLM" in capital letters. Godot's June 2026 policy blocks new contributors with fewer than three merged PRs from submitting features without maintainer permission, bans autonomous agents and vibe coding, and requires disclosure for mechanical AI use. System76's COSMIC desktop now requires contributors to check a box declaring no AI-generated code, comments, or descriptions.
RedMonk analyst kate holterhoff mapped 86 organizations into three stances: ban (Gentoo, QEMU, Zig, Godot), disclose (Linux kernel, LLVM, Ghostty), and allow with responsibility (curl, CPython, Node.js). The trend is toward stricter policies, with Godot and System76 acting in late 2026. A Linux Foundation survey estimates about 25 percent of new PRs are low-quality AI slop.
The cost asymmetry between cheap AI-generated PRs and expensive human review is the root driver of the bans; System76's Jeremy Soller frames it as pure arithmetic: near-zero submission cost, unchanged review capacity.
The top HN joke notes that a site cataloguing anti-vibe-coding projects may itself have been vibe-coded, but the real target is contribution quality, not tool use alone.
Policy movement is clearly from allow to disclose to ban, with late-2026 actions from Godot and System76 marking an inflection point.
Detection remains unreliable, so bans depend on self-disclosure and human review, creating a real risk of false positives for developers who use legitimate AI assistance like autocomplete.
What's even more interesting is that this article of yours was also written with AI—using an AI-written article to discuss a vibe-coded website that catalogs projects opposing vibe coding.