The GNOME Shell Extensions store has updated its guidelines to prohibit AI-generated extensions amid a surge in low-quality submissions. Developers may still use AI as a tool for learning and development, but code primarily written by AI will be rejected. This move aims to maintain code quality and reduce review delays.
Earlier this month, the GNOME Shell Extensions store revised its review guidelines to explicitly state that extensions must not be AI-generated, as noted by reports from It's FOSS and Phoronix. This change addresses a growing influx of submissions featuring poorly written code produced by AI tools, often submitted by developers who do not fully understand the output.
Javad Rahmatzadeh, a developer who reviews extensions for GNOME, detailed the issue in a blog post. He highlighted that the use of AI has resulted in packages containing many unnecessary lines and bad practices. Rahmatzadeh spends up to six hours a day reviewing over 15,000 lines of code and engaging with the community. In the past two months, the number of new extensions has increased, which is positive for community growth, but problematic submissions have prolonged review times. As he explained: "This has led to receiving packages with many unnecessary lines and bad practices. And once a bad practice is introduced in one package, it can create a domino effect, appearing on other extensions."
GNOME extensions perform various functions, including essential customizations like the popular Dash to Dock, which enhances control over the app dock. The guidelines permit AI use as a learning aid or development tool, provided developers comprehend the code they submit. Earlier this year, the Fedora Council adopted similar measures against so-called 'vibe coding' with AI, though without a full ban.
This policy seeks to ensure high standards in the Linux desktop environment, preventing the spread of subpar code that could affect user experience and extension reliability.