AI NewsOpen sourceReported
System76's COSMIC desktop bans AI-written code and docs from contributions
System76 added a no-AI rule to the COSMIC desktop's contributor guide, while a GNOME developer argues the project must accept AI-written bug reports to keep up with the flaws AI is now finding in older code.

Image: The Register
Why it mattersA desktop a reader may install is now rejecting every pull request that used an AI assistant, so a contributor has to say in the PR itself whether a model touched any part of the submission.
A contributor who used Claude or Copilot to write a line of code for the COSMIC desktop can no longer send that code in a pull request. System76, the company behind COSMIC, added a rule to the project's contributor guide that asks every contributor to confirm, in the pull request itself, that no large language model wrote any of the code, the comments, the documentation or the submission text. Liam Proven at The Register reported the change on 7 October.
What the rule says
The exact line the guide now asks a contributor to tick reads: "I have not included any LLM (also known as AI) generated content in this PR, including code, comments, and descriptions." A contributor may still use AI to study the project or to find a bug, but the submitted work has to be theirs.
Two GNOME sub-projects already carry similar rules. The GNOME Calendar contributor guide says the project "does not allow contributions generated by large languages [sic] models (LLMs) and chatbots." GNOME Extensions permits a contributor to use AI for learning and for code completion, but its review rule is that extensions "must not be AI-generated."
GNOME is debating the other direction
Michael Catanzaro, a GNOME developer, published a post titled "The Era of Software Quality, or the Era of Ostriches?" on 2 October, arguing that GNOME must accept AI-generated bug reports. It follows a June post of his, "Please Do Not Ban AI-Assisted Issue Reports." He writes that GNOME is largely written in C, with parts in C++, Vala and JavaScript, and that experienced developers cannot write secure code in memory-unsafe languages for long. He points to the 1,313 CVE identifiers in Debian's latest kernel security advisory as the scale of what is now being found.
In July, Catanzaro cut GNOME Security's disclosure deadline from 90 days to 30, effective 1 August. The Register notes that Red Hat employs many GNOME contributors, that Red Hat's management has been publicly enthusiastic about AI, and that letting AI-generated reports in opens a door that may not stop at reports.
Other projects have already picked a side
Debian decided to allow AI-assisted contributions, and the Linux kernel does too, both cited in The Register's piece. The KDE project is debating it. The COSMIC decision puts a reader's desktop on the stricter side of the question, while the GNOME debate shows that the same organisation can hold both positions in different sub-projects at once.
A contributor to any of these projects has to read the contributor guide before the next pull request, because the rule that applied in July may not apply in October. A review that lands on a bot-written patch in a project that bans them is a wasted submission, and a report that a human never would have found is now in the pool for maintainers whose project accepts them.
Source
COSMIC shuts the door on AI code as GNOME debates letting bug reports in, The Register, Liam Proven.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.