COSMIC shuts the door on AI code as GNOME debates letting bug reports in
The Register
System76 has banned AI-written code and docs from COSMIC contributions. GNOME is still arguing over whether AI bug reports should be welcome.
Based on reporting by The Register — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
System76 has drawn a hard line around its COSMIC desktop: if the text, code, comments, or pull request notes were written by an LLM, the contribution stays out. The project’s updated contributor rules now require people to say they did not include any AI-generated material in their PRs. You can use AI to learn, and apparently to hunt for bugs. You just can’t hand the project AI-written work.
That puts COSMIC in the same camp as parts of GNOME, which has its own restrictions on AI-generated submissions. GNOME Calendar says it does not allow contributions generated by large language models and chatbots, while GNOME Extensions says extensions must not be AI-generated. It does still allow AI learning tools and code completion, which is a more permissive line than COSMIC has taken.
The timing matters because COSMIC is still young, even if the project itself is not. The Register first wrote about it nearly five years ago, before ChatGPT arrived in late 2022. The desktop’s first stable release shipped with Pop!_OS 24.04 in late 2025. It looks GNOME-ish — top panel, dock-like launcher — but the new COSMIC is a ground-up Rust project using the Iced toolkit and runs on Wayland only.
GNOME is the older giant here, not the newcomer. It dates back to 1997, landed its first default Debian-era-style foothold at Red Hat in 1999, and is still the default desktop in RHEL. Much of it is written in C, with GNOME Shell mixing JavaScript and C, and some pieces in Vala. That matters because Michael Catanzaro, a GNOME developer, is arguing the project should accept AI-generated bug reports, saying secure code is hard to write in unsafe languages like C, C++, and Vala.
His case is not subtle: AI is already surfacing huge numbers of bugs in huge FOSS codebases, and refusing those reports may just leave problems sitting there. But once a project opens the door to AI-written reports, the next arguments practically write themselves: triage, fixes, extensions, core code. Red Hat’s general enthusiasm for AI makes that path feel less hypothetical than it should.
My take — AI-written commentary, not fact-checked reporting
This is the part where open source discovers that “AI allowed for reports, banned for code” is a temporary truce, not a philosophy. COSMIC is doing the sensible thing by keeping the patch stack human, while GNOME is staring at the bug queue and reaching for the first shovel it can find. That is exactly how standards drift: one exception at a time, each one sold as harmless, until the room is full of exceptions.
Read more about this at: The Register
Related stories
Linus Torvalds to critics of AI coding in Linux: "Fork it. Or just walk away."
Ars Technica · 2 months ago ·
24