Pop!_OS bans AI-generated code, comments and descriptions from its projects: Reveneau

Why it mattersA contributor who was told to disclose AI use now has to not use it at all, and a project that accepts automated patches today may close them with no comment tomorrow.

A contributor who sent a tested patch to Pop!_OS this week got the pull request closed the moment they wrote that an LLM had helped them find and fix the bug. The clause that now does the work was added to CONTRIBUTING.md on 30 September and tightened the next day.

System76, the Denver computer maker that develops Pop!_OS and the COSMIC desktop, added an "AI Contributions" section to its CONTRIBUTING.md on 30 September 2026. The commit message was "CONTRIBUTING.md: add brief AI/LLM policy" and the sign-off is Jeremy Soller, who is both the author of the commit and System76's principal engineer. The next day Soller shortened the clause further, with a commit note stating that listing acceptable AI uses "may be interpreted as condoning those usages, when it does not".

What the policy says

The current text in the Pop!_OS repository reads, in full: "Content generated by any LLM (also known as AI) may not be contributed in issues or PRs to Pop!_OS projects including code, comments, and descriptions." The 30 September version carried a second sentence allowing LLMs for "research, testing, benchmarking, and other tasks". That sentence was removed in commit ce5545a on 1 October.

The pull request template in pop-os/cosmic-epoch, the COSMIC desktop repository, now carries a matching checkbox: "I have not included any LLM (also known as AI) generated content in this PR, including code, comments, and descriptions." The older template, in use until the last days of September, asked contributors to disclose AI use and warned that "AI authored or partially authored PRs may be closed without comment".

Who it covers, and the effect on existing work

The clause applies to every Pop!_OS repository under the pop-os GitHub organisation, which includes pop-os/cosmic-epoch and the dozen or so COSMIC crates it tracks as submodules. A contributor's commentary in a Hacker News thread the same day described having a password-handling fix for the Pop!_OS network applet closed by a maintainer after they disclosed using Claude to help identify and fix the bug. The author wrote that the quality was handcrafted but did not contest the decision.

The policy forbids AI-generated text and code reaching the project through an issue or a pull request, and says nothing about using AI for other work in private: finding a bug, running a test, benchmarking a build. In practice an issue or a pull request is the gate every outside contribution passes through.

What a maintainer or a contributor does on Monday

If you have an outstanding Pop!_OS pull request where you used an LLM to draft code, comments or the description itself, the maintainer is now entitled to close it with no comment under the new template. Rewriting those parts by hand, with an explanation in the commit of what was reconsidered, is the only path that fits the policy as written. If you maintain a project that depends on COSMIC and takes AI-assisted patches of your own, there is no requirement to match this stance, and the Linux kernel's own position is that AI-assisted contributions are permitted when the author takes responsibility for them. A desktop project refusing AI code while its upstream accepts it is now a common split.

Source

System76, CONTRIBUTING.md, current version, retrieved 2026-10-03.
Jeremy Soller, "CONTRIBUTING.md: add brief AI/LLM policy", commit, 30 September 2026.
Jeremy Soller, "Do not explicitly provide acceptable LLM usage", commit, 1 October 2026.
Hacker News, discussion of the policy, 3 October 2026.

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.