Open-source projects that refuse AI-generated contributions: what contributors and maintainers should keep

Flat illustration of a pull-request page with an orange stop badge beside a log card, with a clock and a teal check-mark badge. No people, text or logos.

Short answer: A community tracker called osssaysno.com listed 22 open-source projects on 11 November 2026 that refuse AI-generated contributions, including QEMU, Gentoo, Zig, Servo and immich, according to a FreakyJolly summary. The reasons given are unclear copyright status of AI output, the Developer Certificate of Origin (DCO), reviewer overload and quality. Other projects, such as the Linux kernel, allow AI help if a human takes responsibility. Before you send a patch, read the project’s policy and keep your own dated working record. A TimeBind dates the files you hold. It does not show whether any line was written by a person or a tool.

Last checked 13 November 2026. General information, not legal or security advice. The project list below comes from a third-party summary dated 11 November 2026 of a community tracker. Policies change, so read each project’s own contributing file before acting.

Table of Contents

What the tracker and its summary say

FreakyJolly’s 11 November 2026 write-up says the tracker had 22 projects, each with a link to its policy and a “verified” date, and that the list was refreshed that day. The projects named include QEMU, Gentoo Linux, Zig, Servo, Redox OS, immich, Meilisearch, libvips, MonoGame and stb. The summary groups the stated reasons into three: unclear copyright or licence status of AI output, reviewer overload from machine-generated pull requests, and quality or trust. It says QEMU describes the copyright and licence status of AI output as “ill-defined” and that it is “not willing or able to accept the legal risks”, and that contributors sign a Developer Certificate of Origin they cannot honestly sign for AI output.

It also notes the other side: the Linux kernel and Fedora allow AI assistance if a human takes responsibility and discloses it. Policies differ in scope too. Some cover only code, others cover documentation, issues and art.

A maintainer’s view: OpenLDAP

OpenLDAP’s Howard Chu has repeated in public posts that the project accepts contributions only from the original author and that AI-generated contributions must be rejected because their provenance is unknown. The Lunduke Journal post sharing the claim on 9 November 2026 drew 6,346 impressions and 222 likes when we measured it on 13 November. That is one maintainer’s stated policy and reasoning, not a legal conclusion. Whether any given AI output infringes anything is a legal question these pages do not answer.

What contributors should do and keep

  1. Read the project’s contributing file and any AI policy first. Some projects close or ban for a largely AI-generated pull request, and several punish hiding AI use harder than using it.
  2. Keep your working record before you open the pull request: your branch history, your local notes and design sketches, and the date you started. A repository’s own history is a good record, but you control a copy outside it too.
  3. If the project allows AI help, keep the tool, the date and a summary of what it did, in the form the policy asks for.
  4. Fingerprint a snapshot of your work before you submit. A TimeBind of the folder gives you a dated record that those exact files existed no later than the recorded time and still match. See how to prove code existed at a specific time and proving source code existed before a date.

The record does not show how the code was written. It supports the narrower claim that this version existed by that time and has not changed since.

What maintainers should keep

If you set or change a policy, keep a dated copy of each version and of the discussion that led to it, so contributors and downstream users can see which rules applied when. The open-source licence change record checklist covers the same habit for licence changes, and the Asahi Linux relicensing piece shows why dated records matter when a project changes its terms.

Why people are searching for this

Measured on 13 November 2026: the OpenLDAP post above had 6,346 impressions and 222 likes. The tracker appeared on Hacker News, and FreakyJolly reported it went viral on 11 November. A video about Asahi Linux’s relicensing had 77,596 views on YouTube. These are separate signals about the same debate, not one count.

Frequently Asked Questions

  1. Which projects refuse AI-generated contributions?

    As of 11 November 2026, the osssaysno.com tracker listed 22, including QEMU, Gentoo, Zig, Servo and immich, according to FreakyJolly. Check each project’s own policy page, because the tracker is a community summary and policies change.

  2. Why do some projects refuse them?

    Reasons quoted in the summary include unclear copyright and licence status of AI output, difficulty honestly signing the Developer Certificate of Origin, reviewer overload and quality concerns.

  3. Do all projects ban AI help?

    No. The summary notes that the Linux kernel and Fedora allow AI help if a human takes responsibility and discloses it.

  4. Can a TimeBind show my code was not written by AI?

    No. A TimeBind is a dated, checkable record that an exact file existed no later than a recorded time and still matches. It records timing and integrity, not how the code was made.

  5. What should I keep before sending a pull request?

    Your branch history, notes and design files with dates, any disclosure the project asks for, and a fingerprinted snapshot of the work as you submit it.

Sources checked

Date your working files before you submit a patch

Your first Pooled TimeBind is free, with no credit card required. After that, TimeBinds start from US$5, and an ASAP TimeBind is US$15. Click here to start, or see plans and pricing.