Committee Approves New Rules for GNOME Circle Directory
The committee responsible for managing the GNOME Circle directory has recently approved new rules aimed at prohibiting the publication of applications created using AI tools. GNOME Circle serves as a platform for third-party developers to showcase applications and libraries developed using GNOME technologies, providing them with a gateway into the GNOME ecosystem.
Under the new rules, GNOME Circle will no longer accept projects that exhibit indications of being generated using AI, such as the presence of meaningless code insertions, inconsistencies in coding style, contrived use of APIs, and comments containing hints for AI. This decision was made in order to accelerate the mandatory review process for projects seeking inclusion in the GNOME Circle collection. Due to a substantial backlog in the review queue, some applications have had to wait for years before a decision was made regarding their inclusion.
The recent surge in the submission of low-quality AI-generated programs has added to the workload of reviewers and contributed to further congestion in the application queue. In an effort to address this backlog, the acceptance of new applications for review has been temporarily suspended. Once this suspension is lifted to allocate additional resources, any projects created using AI will be automatically rejected without undergoing the review process (with the exception of applications submitted prior to the implementation of the new rules).
Another rationale for the ban on AI usage is the enforcement of stricter guidelines within the Flathub directory, which houses software packages featured in the GNOME Circle. As Flathub’s regulations indirectly affect the GNOME Circle project, it is deemed unnecessary to invest time in reviewing programs whose packages would not meet Flathub’s standards.
A survey conducted among GNOME Circle maintainers revealed that 62% of participants do not utilize AI in their development process, 34% rely on AI assistants for minor tasks or to generate basic code snippets, while only 3% utilize AI for generating substantial portions of code. Furthermore, none of the maintainers admitted to completely replacing hand-written code with AI-generated code.