Generative AI Policy - Asahi Linux Documentation
It is the opinion of the Board that Large Language Models (LLMs), herein referred to as Slop Generators, are unsuitable for use as software engineering tools, particularly in the Free and Open Source Software movement.
The use of Slop Generators in any contribution to the Asahi Linux project is expressly forbidden. Their use in any material capacity where code, documentation, engineering decisions, etc. are largely created with the "help" of a Slop Generators will be met with a single warning. Subsequent disregard for this policy will be met with an immediate and permanent ban from the Asahi Linux project and all associated spaces.
All of the popular Slop Generators are trained on an incomprehensibly large corpus of text. There is ample evidence across the Web of this training material including copyrighted material, brazenly stolen by the Slop Generator proprietors with impunity. Due to the nature of Slop Generators, they are prone to regurgitating their training corpus almost verbatim. This presents a challenge for FOSS projects in that the use of generated slop is highly likely to violate intellectual property law by way of regurgitating the aforementioned stolen training material. This likelihood is proportional to the specificity of the problem area.
Asahi Linux is a highly specific project, working in esoteric problem spaces on publicly undocumented hardware. Given the techniques used by Slop Generator manufacturers, it is not impossible for them to have confidential or leaked material owned by Apple or its vendor partners in their training corpi. It is therefore likely that Slop Generators will regurgitate this when queried in just the right way. We already forbid the use of illegally acquired or leaked documentation and tooling (e.g. Apple's internal repair diagnostic tools). This also applies to regurgitated slop.
%20(1).png)
