Hello Guix! I am submitting GCD 008: “Standing up for human crafting” (draft attached).
Two fellows kindly offered to sponsor it already; I’ll wait for a couple of days before submitting the draft on Codeberg in case others want to sponsor it. The “discussion period” will formally start once it’s on Codeberg. Thanks in advance, Ludo’.
title: Standing up for human crafting id: 008 status: draft discussion: https://issues.guix.gnu.org/<number assigned by issue tracker> authors: Ludovic Courtès sponsors: Janneke Nieuwenhuizen, Noé Lopez date: <date when the discussion period starts> draft-date: <date when the draft is posted to look for sponsors> discussion-date: <date when the discussion period starts> deliberation-date: <date when the deliberation starts> SPDX-License-Identifier: CC-BY-SA-4.0 OR GFDL-1.3-no-invariants-or-later --- # Summary Guix is a free software project; it is made by people for people, with knowledge sharing and empowerment as some of its core values. This document proposes a clear stance and a policy to support this vision. # Motivation Free software as a project is fundamentally about humans: it is about producing software that guarantees the freedom of its users, and in particular freedom from the original developers; it is about empowering people, increasing their autonomy, by providing them the means to study and adapt the software to their needs; it is about knowledge sharing and mutual aid. Through this humanist approach, thousands of people have been producing source code that is now used as raw material for stochastic text generators—large language models (LLMs), also referred to as “generative artificial intelligence” or “genAI”—primarily backed by large corporations. However impressive the results may look, the authors believe genAI has a social and environmental impact that undermines the very humanist foundations of free software and Guix: - The leading commercial genAI services run non-free software, are a threat to user privacy, and achieve a concentration of power over the lives of people rarely seen before. - GenAI launders the reciprocity baked into copyleft licenses such as the GNU General Public License (GPL), effectively violating it. A real-world example of copyleft-laundering is [the `chardet` LLM-assisted “rewrite” for the stated purpose of relicensing from LGPL to MIT/Expat in March 2026](https://tuananh.net/2026/03/05/relicensing-with-ai-assisted-rewrite/) ([covered by LWN](https://lwn.net/Articles/1061534/)) or the [EmDash WordPress reimplentation in TypeScript “under the more permissive MIT license”](https://blog.cloudflare.com/emdash-wordpress/). - The narrative around the output of genAI tends to demean human creativity, to [discourage the learning and knowledge sharing processes](https://cekrem.github.io/posts/programming-as-theory-building-part-ii) at the core of free software, and to [discourage free software contributions by humans](https://arxiv.org/pdf/2601.15494). We believe it [stifles individual autonomy](https://ali-alkhatib.com/blog/defining-ai) at a fundamental level—replacing one’s ability to build up knowledge with a false sense of quick achievement, building up [cognitive debt](https://simonwillison.net/2026/Feb/15/cognitive-debt/)—while also [weakening communities and destroying labor power](https://tante.cc/2026/04/21/ai-as-a-fascist-artifact/). - The huge ecological footprint of genAI is well documented, going from the [energy](https://cleanview.co/content/power-strategies-report), [water, and materials needed to build the servers used to train models and to service requests](https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/), to the land usage of those data center, and to the energy and [water needed to operate those data centers](https://theconversation.com/data-centers-consume-massive-amounts-of-water-companies-rarely-tell-the-public-exactly-how-much-262901). Fair sharing of natural resources and the habitability of the planet are prime concerns from a humanist perspective. To this humanist perspective, we should add legal concerns: - At the time of writing, only proposed interpretations of copyright law exist: that depending on the level of human intervention, genAI output could be considered not copyrightable or at best “uncertain” in the [European Union](https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf) (p. 99–103) and in the [USA](https://www.congress.gov/crs-product/LSB10922) (see also this US Copyright Office [report on copyrightability](https://copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf) and the White House March 2026 [recommendations](https://www.whitehouse.gov/wp-content/uploads/2026/03/03.20.26-National-Policy-Framework-for-Artificial-Intelligence-Legislative-Recommendations.pdf) acknowledging that this is unsettled.) - Given the unsettled situation of copyright on LLM output at an international level, incorporating LLM output in a free software project is risky. In the Chardet case mentioned above, the Software Freedom Conservancy (SFC) [calls on everyone to *not* assume its relicising is legitimate](https://github.com/chardet/chardet/issues/355#issuecomment-4145369025). - This legal uncertainty is one reason for projects [such as Gnulib](https://lists.gnu.org/archive/html/bug-gnulib/2026-02/msg00064.html) to prohibit the inclusion of [“legally significant”](https://www.gnu.org/prep/maintain/html_node/Legally-Significant.html) portions of code (more than 10 lines). - LLMs currently used in production by commercial services are known to [regurgitate entire parts of their training data verbatim](https://arxiv.org/abs/2601.02671), thereby infringing on copyright of the original works. - A [class action against Anthropic](https://www.anthropiccopyrightsettlement.com/) is underway in the USA, for alleged copyright infringement of some of its training data (it is being settled without trial though). - In France, a [proposed law](https://www.senat.fr/leg/ppl25-220.html) would, upon litigation, require genAI providers to prove that the works at hand were not part of their training data (rather than the other way around), though this is currently limited to cultural work. This document proposes the adoption of a pledge in support of human production. We are well aware that commercial genAI services perform well on tasks relevant to Guix, first and foremost packaging. There is no doubt that genAI is already being used within the community. Our goal is not to judge what individuals are doing. Instead, this proposal aims at setting a standard for what we do collectively within the project. # Detailed Design We propose the adoption of a pledge on behalf of the project, with the following goals: - Contributing to the public debate on these matters and creating ties with like-minded organizations and grassroots movements. - Strengthening support for craftspeople the project interacts with—translators, artists, developers, and so on. - Questioning the reasons that make genAI feel necessary for people using Guix, and finding ways to fill the gap. This pledge is *not* about claiming a “moral superiority” of the project and its members for not resorting to genAI, nor is it about “virtue signaling”. The authors acknowledge that genAI is already widely used, is often even hard to escape, and creates a dependency similar to other dopamine-inducing processes. Consequently, our goal is to help people make do without genAI, not to point fingers at individuals. ## Pledge We propose the following project commitments: 1. The project **will not use nor encourage use of genAI** for its code, packages, code review, artwork, translations, or any other artifacts. 2. We kindly ask contributors to respect this choice and not use LLMs for their contributions to Guix. Nevertheless, code claimed to be produced in whole or in part by genAI **may be incorporated in the limit of at most 15 lines of code** to ensure the contributor has a valid copyright claim on the code. 3. Software where the majority of commits were authored or co-authored by genAI **will not be packaged in Guix**. Notable examples of such code include [Claude’s C compiler](https://github.com/anthropics/claudes-c-compiler/), [EmDash](https://github.com/emdash-cms/emdash), and [Neomacs](https://github.com/eval-exec/neomacs), 4. Packages in Guix will always be **built from source**, the only exceptions being compilers or build systems for which a bootstrap has yet to be found (a notable example is GHC). For software that includes neural networks, we consider the Corresponding Source to include all the training data; software for which training data is unavailable, or for which re-computing weights from training data is infeasible, *cannot be included in Guix*. For example, LibreTranslate, which [downloads models for local use](https://lists.gnu.org/archive/html/help-guix/2026-04/msg00004.html), may not be included in Guix. Conversely, GNU Backgammon, [which can recompute its neural network weights](https://mastodon.nz/@gtw/115851324313436125), is acceptable for inclusion in Guix. Should applying this rule lead to package removal, the [Deprecation Policy](https://guix.gnu.org/manual/devel/en/html_node/Deprecation-Policy.html) must be followed. 5. The project will work to **provide people of all levels of experience with the resources to use Guix and to contribute to Guix** without feeling the need to resort to genAI: - by facilitating access to documentation and improving it to better suit the needs to people with different levels of experience—newcomers discovering free software, people with prior exposure to GNU/Linux, developers; - by developing and improving tools that make it easier to get started such as [Guix Packager](https://guix-hpc.gitlabpages.inria.fr/guix-packager/) and `guix import`; - by improving diagnostics to make our tools more approachable; - by providing communication channels anyone can use to look for help or to offer support, where all and everyone can feel safe to participate; - by improving tools for mechanical translation, refactoring, and updates such as `guix import`, `guix style`, and `guix refresh`, so that tedious packaging tasks can be automated in a way that is transparent, deterministic, and maintains the project’s sovereignty. 6. We acknowledge that the project’s sustainability depends on automation for all the mechanical, labor-intensive tasks such as package updates. We will keep **improving tools and services to automate some of the package collection maintenance work**. The Guix project will publicize this pledge, by referring to it in its contribution guidelines, by making it easy to find, by presenting it in blog posts, and through any other communication deemed appropriate. ## Related Work A number of free software projects have adopted a policy with respect to genAI, revealing different sensibilities and choices. Here are some of those we looked at: * Asahi Linux https://asahilinux.org/docs/project/policies/slop/ * Debian https://lwn.net/Articles/1020968/ (May 2025) * FreeBSD https://www.heise.de/en/news/FreeBSD-policy-AI-generated-source-code-No-thanks-10634141.html * Gentoo https://wiki.gentoo.org/wiki/Project:Council/AI_policy * GNOME (individual packages) https://gitlab.gnome.org/GNOME/gnome-calendar/-/merge_requests/725 https://discourse.gnome.org/t/loupe-no-longer-allows-generative-ai-contributions/27327 https://gitlab.gnome.org/GNOME/libadwaita/-/blob/main/CONTRIBUTING.md#use-of-generative-ai * GNU Binutils https://sourceware.org/binutils/wiki/LLM_Generated_Content * GNU Gnulib https://lists.gnu.org/archive/html/bug-gnulib/2026-02/msg00064.html * Linux https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/generated-content.rst https://lwn.net/Articles/1032612/ https://docs.kernel.org/process/coding-assistants.html * LLVM https://github.com/llvm/llvm-project/blob/main/llvm/docs/AIToolPolicy.md * Mastodon https://github.com/mastodon/.github/blob/main/AI_POLICY.md * Pepper & Carrot (comics) https://www.peppercarrot.com/en/documentation/409_Code_of_Conduct.html#use-of-generative-ai * Servo https://web.archive.org/web/20260311122512/https://book.servo.org/contributing/getting-started.html#ai-contributions * Zig https://ziglang.org/code-of-conduct/#strict-no-llm-no-ai-policy Software package lists: * "No AI" list https://noai.starlightnet.work/list.html * software projects that accept LLM contributions https://codeberg.org/small-hack/open-slopware ## Cost of Reverting Should consensus be found *via* a new GCD, the pledge could be removed or amended. # Drawbacks and Open Issues This proposal takes a clear stance that not everyone may agree with. This could lead to fragmentation within the Guix community, or within the free software community.
signature.asc
Description: PGP signature
