AI is forcing open-source projects to define what “contribution” really means. COSMIC has drawn a firm boundary around AI-generated code and content, while GNOME is debating whether AI-assisted bug reports belong inside its project.
On 7 Oct 2026, Liam Proven published an article examining the split. One project is putting strict limits on machine-generated contributions, while the other is asking whether AI can help developers report problems without weakening the quality of the project’s work.
COSMIC Puts Human Disclosure at the Center
COSMIC, the Linux desktop environment developed by System76, bans AI-generated content from contributions to the desktop. The latest version includes a rule requiring contributors to declare that they have not included any LLM-generated content in a pull request.
The declaration covers more than source code. Contributors must state: “I have not included any LLM (also known as AI) generated content in this PR, including code, comments, and descriptions.” That wording reaches into every part of a contribution, from the implementation itself to the explanation attached to it.
The policy gives COSMIC a clear answer to a question spreading across software projects: can a contribution remain fully human-authored when an AI system writes even part of the code, comments, or description? For COSMIC, the answer is no under this rule.
COSMIC is one of the newest complete Linux desktop environments, and System76 has worked on it for years. Its first stable release shipped with Pop!_OS 24.04 in late 2025, marking a major step for a desktop that began as a customized direction and grew into its own environment.
The technical foundation also sets COSMIC apart. It is written in Rust using the Iced toolkit and is Wayland-only. The original System76 desktop called “COSMIC” was based on GNOME and heavily customized with GNOME extensions, but the newer COSMIC represents a separate path for the Linux desktop.
GNOME Keeps AI Out of Code, But Questions Bug Reports
GNOME has its own restrictions on AI-generated development work. Its Calendar guide to contributing states: “This project does not allow contributions generated by large languages [sic] models (LLMs) and chatbots.” The GNOME Extensions review guidelines add an even shorter rule: “Extensions must not be AI-generated.”
Those rules draw a line around submitted code and extensions, but GNOME developer Michael Catanzaro is asking whether the same line should apply to bug reports. Catanzaro published a blog post advocating for the acceptance of AI-generated bug reports, opening a new debate over whether issue reports should follow the same standards as code.
Catanzaro linked to an earlier post from June titled “Please Do Not Ban AI-Assisted Issue Reports.” The argument creates a distinction between using AI to create software and using AI to describe a problem. A bug report does not change the project in the same way as code, yet its quality still affects the work developers must do next.
Catanzaro framed the larger question with another title: “The Era of Software Quality, or the Era of Ostriches?” That question points directly at the pressure facing open-source maintainers. Rejecting every form of AI assistance may protect a project’s standards, but accepting machine-generated reports may help more problems reach developers.
The debate also arrives as GNOME continues a long technical history. The GNOME project was announced in 1997, GNOME 1.0 was released in 1999, and Red Hat Linux included a preview of GNOME in 1998. GNOME 1.0 became the default desktop in Red Hat Linux 6.0 in 1999, and GNOME remains the default desktop in RHEL.
Two Different Rules for an AI-Filled Development Era
GNOME is built in C using GTK, with much of its code written in C. GNOME Shell combines JavaScript and C, and some GNOME components are written in Vala. That mix shows how a mature desktop project can carry several programming languages while still maintaining rules for how new contributions enter the codebase.
Security adds another layer to the discussion. Michael Catanzaro announced a reduction in the disclosure deadline for issues reported to GNOME Security from 90 days to 30 days, effective August 1. At the same time, Debian’s latest Linux kernel security advisory listed 1,313 CVE identifiers, underscoring the scale of the software problems that developers and maintainers must track.
Against that backdrop, the difference between COSMIC and GNOME becomes easy to see. COSMIC requires contributors to declare that their pull requests contain no LLM-generated content, including code, comments, and descriptions. GNOME rejects AI-generated project contributions and extensions, while Catanzaro argues that AI-assisted issue reports deserve consideration.
Neither position removes the central challenge. Open-source projects need useful code, clear reports, and trust between contributors and maintainers. AI can touch each of those areas, so rules that once focused on authorship now reach into review, disclosure, security, and the basic meaning of a contribution.
COSMIC’s policy makes human authorship a condition of entry. GNOME’s debate leaves more room for AI-assisted reporting while keeping restrictions on generated project work. As both desktops continue to evolve, those choices will shape how developers build, review, and protect the Linux environments they depend on.
