Changes to Sourcehut's terms of service regarding LLMs

SourceHut has decided to update its terms of service to restrict or prohibit the use of large language models (LLMs) and other generative AI technologies on its platform, following a community discussion and internal deliberations. The decision, announced by SourceHut staff, is driven by concerns over resource usage, outages, and ethical considerations, with the community broadly supporting the restrictions. The changes will be implemented with specific enforcement examples and a timeline, and options for affected projects will be provided.

In late July, Codeberg announced that they would limit the use of LLMs on their platform https://blog.codeberg.org/protecting-our-floss-commons-from-llms.html , something that SourceHut has also been considering for some time. Last week, we discussed with the SourceHut community https://lists.sr.ht/~sircmpwn/sr.ht-discuss/%3CDKSTMKM0ZD9N.2FTBDFREZH699@ddevault.org%3E the possibility of limiting or prohibiting the use of LLMs large language models and other generative AI technologies on SourceHut. Thank you to everyone who participated, shared a variety of views, and ensured we had a productive and civil conversation on the subject. Following this conversation and more discussions internally, we have elected to move forward with the necessary changes to our terms of service. I encourage you to read through this thread to see the varied and thoughtful opinions of the SourceHut community on this subject, both for and against the change. In summary, the SourceHut community broadly has strong support for restrictions on the use of AI on this platform. However, ultimately this decision was made by staff, so we’ll focus on our own rationale for it, give examples of how the new policies will be interpreted and enforced, the expected timeline for the change, and options for people affected by the change and how they can move their projects forward afterwards. = Click this to skip the background and read about the changes changes There are material incentives for us to curtail “vibe coded” projects on SourceHut. They tend to, as a class, be outliers in terms of resource usage, as the practice tends to invent complex CI configurations that waste build minutes, produce larger codebases, and push commits often, all of which causes excessive resource usage and can lead to outages. Often these codebases are abandoned after burning through all of these resources for naught. And, of course, the scourge of LLM crawlers scouring the internet causes more outages and headaches for our sysadmins, and AI is causing skyrocketing costs for the hardware we need to run the platform and the hardware you need to use it. But, in truth, the rationale is more about politics and ethics than anything else. Deciding which “hills to die on” is tricky. We are an organization which is transparent about politics, both the fact that we, like all others, have politics, and also about what our politics are. But, although the world is a turbulent place, and this calls all of us to action, there are many issues we believe firmly in but which need not necessarily steer SourceHut policy. If we were to shape our policies in terms of our values on, for example, Palestine, it would serve to stoke controversy and increase our workload without meaningfully contributing anything to the problem. We lack the cultural and intellectual context to know what to do about that situation, and we have access to few resources knowledgeable enough to steer our policies with wisdom and care. In the case of AI, though it touches on complex social, economic, and political factors, many of which are beyond the realm of our our expertise, it overlaps considerably with matters on which we are experts – software development, open source, and open culture, for example. As experts, we can more easily see that we have an obligation to evaluate their impact and to weigh in and make decisions accordingly. If we focus narrowly on how LLMs affect our community, we can consult our mission statement for guidance: We are here to make free software better. We will be honest, transparent, and empathetic. We care for our users, and we will not exploit them, and we hope that they will reward our care and diligence with success. It’s not easy to argue that LLMs make free software better. The means by which they operate involves taking, by force https://codeberg.org/forgejo/discussions/issues/320 and without regard for the wishes of the software authors, the obligations of software licenses https://spectrum.ieee.org/ai-code-generation-ownership , or the platforms on which our work relies, vast swaths of open source software, disregarding the licensing requirements, and charging rent to produce code which is a plausible reinterpretation of its training data applied to new problems. The models which are produced in this process are proprietary, owned and controlled by private corporations, and give little back to the open source community. Therefore, the relationship between LLMs and open source seems to exploit the users we are responsible for caring for. We have also seen some other troubling incidents, for example the use of LLMs to rewrite open source software to circumvent copyleft obligations, such as in the case of the chardet library https://github.com/chardet/chardet/issues/327 , a practice which seems to straightforwardly work against the interests of open source software authors. However, LLMs have shown some utility. They find real bugs, including security bugs https://lwn.net/Articles/1070698/ , and open source software has become more robust as a consequence. Some programmers find them useful for code reviews https://lwn.net/Articles/1063292/ and other creative applications, and indeed many programmers report good results in using them to write or assist in writing code. The more we zoom out, though, the more alarming it gets. The biggest concern is the impact on the climate. Cars fueled by fossil fuels are also useful, but most people support efficiency improvements and ultimately a complete replacement with electric vehicles to mitigate their effects. The transformation of society to address the climate emergency must be comprehensive, and each of us, in our respective industries and expertise, have a moral obligation to work to reduce our environmental footprint, particularly in systemic and institutional ways. The roll-out of AI is a massive trend in the wrong direction. LLMs and similar tools are an extremely energy-inefficient https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117 way of solving the problems they address, such as authoring software, and are fundamentally at odds with mitigating the climate disaster https://www.politico.eu/article/europe-choose-ai-climate-goals-data-center-chief-warns/ . Planned developments in AI infrastructure are expected to generate an electricity demand comparable to the entire country of India https://newsroom.ucla.edu/stories/opinion-ai-is-destroying-our-planet-we-must-act , a country which is experiencing tens of thousands of deaths https://www.frontiersin.org/journals/environmental-health/articles/10.3389/fenvh.2026.1789071/full due to heatwaves exacerbated by climate change. This build-out consumes a large portion of the green energy coming onto the grid, delaying de-carbonization of other sectors https://www.reuters.com/technology/artificial-intelligence/how-ai-cloud-computing-may-delay-transition-clean-energy-2024-11-21/ , or else AI drives demand for new fossil fuel infrastructure outright https://www.entergy.com/news/entergy-louisiana-announces-a-new-agreement-with-meta-that-will-deliver-an-additional-2b-in-customer-savings . In either case the average energy consumer often pays the bill https://eelp.law.harvard.edu/extracting-profits-from-the-public-how-utility-ratepayers-are-paying-for-big-techs-power/ . It’s often unclear how to respond to the climate crisis at an individual level. Taking shorter showers or using your air conditioner less might help – and the wealthy tend to highlight these “solutions” to diffuse responsibility for their own impact – but this isn’t comparable to the meaningful institutional change that needs to take place. Individuals do make a difference when they have their hands on the levers of those institutions. As software engineers we have front-row seats for the deployment of one of the world’s fastest growing energy consumers https://energy.ec.europa.eu/news/focus-data-centres-energy-hungry-challenge-2025-11-17 en : AI. Our conscience calls us to resist the deployment of technologies which will exacerbate death and suffering for millions. AI might deliver benefits to society, but the cost is so high that we need to make careful decisions about how it’s deployed and if the benefits to society outweigh those costs. On the whole, vibe coding does not deliver benefits which, in the view of SourceHut and our community, outweigh the costs. Cheap, green electricity, preservation of freshwater resources, husbanding our planet and its wildlife: these are all much more important than AI-assisted coding. The broader social and economic consequences add to the alarm. These tools are disrupting labor and increasing the already unprecedented gap between the haves and the have-nots. These companies pay minimum wage in inhumane conditions in impoverished countries https://time.com/6247678/openai-chatgpt-kenya-workers/ for the labor they exploit to train models that can be applied to taking the bottom out of the labor market in wealthy countries. Moreover, the accumulated wealth of the AI owners, and the depleted resources of the AI proletariat, is further exploited to consolidate power, enable the rise of fascism, and destroy the social order alongside the environmental order. AI is being used to deport immigrants without due process https://www.aclu.org/news/privacy-technology/palantir-deportation-roundup , surveil and police https://www.aclu.org/news/privacy-technology/machine-surveillance-is-being-super-charged-by-large-ai-models poor and racialized https://www.oxjournal.org/predictive-policing-or-predictive-prejudice/ communities, manipulate elections and political discourse, facilitate crimes against humanity at scale https://www.ohchr.org/en/press-releases/2024/04/gaza-un-experts-deplore-use-purported-ai-commit-domicide-gaza-call , and amplify injustices in our institutions. Finally, AI simply seems to be bad for the people who use it. Many open source maintainers complain of being exhausted by a deluge of low-quality AI generated contributions https://arstechnica.com/security/2026/01/overrun-with-ai-slop-curl-scraps-bug-bounties-to-ensure-intact-mental-health/ . AI has been linked to “ deskilling https://arxiv.org/abs/2506.08872 ”, making its users worse at problem solving and other cognitive tasks. And the number of people who have been talked into psychosis https://thehumanlineproject.org/research-stanford/ or suicide https://en.wikipedia.org/wiki/Deaths linked to chatbots Suicides by their chatbots continues to rise. These concerns are myriad, credible, and demand action. Our small contribution to the problem is to clearly state our position on the matter and adjust our policies to provide a safe platform for people who are concious of these issues to work and find community. The upcoming changes The new terms of service, which will go into effect for new projects after the usual two week notice period i.e. September 10th , will now include the following under the list of prohibited content: - original content written with or which facilitates the use of LLMs large language models or other generative AI technologies The use of LLMs or other generative AI tools to produce or assist with the production of source code, assets, tickets, emails, and so on, will no longer be permitted on SourceHut once these terms take effect. As with our other policy changes, the intention is to roll this out slowly and thoughtfully. Each possible violation will be evaluated on a case-by-case basis and we are open to making exceptions. A matter of interpretation This policy is worded carefully to allow some room for reasonable interpretation. For instance, the choice to target “original” content in the policy is intended to allow you to host mirrors of major codebases that may have more lenient LLM policies, such as the Linux kernel. An individual who contributes to a project like Linux can host their personal tree and stage patches on SourceHut, and a Linux kernel subsystem which has an internal policy on LLM use which aligns with SourceHut is welcome to host their tree, mailing lists, and so on, on our platform, without worrying about the impact of AI-assisted code in the rest of Linux. However, SourceHut would not be suitable as the upstream platform for the broader Linux project as a consequence of their liberal LLM policy. Moreover, the use of AI for purposes other than co-authoring code, tickets, emails, and so on, is discouraged but not prohibited. If you use LLMs privately to review your work, to ask questions of, to perform security analysis, and so on, and then write original code to push to SourceHut, your use-case is aligned with our policy. Software which contains generative AI features or directly enables the use of generative AI is not welcome on SourceHut once this policy comes into effect. Enforcement We intend to use a light touch with enforcement. Honesty and transparency is among our core values, and we have extended it to our users and expected it in return. We rely on the honor system when awarding financial aid on the basis of need almost every day, when we refrain from asking users to provide evidence of their financial situation to receive free service. All users pay according to their means, choosing the price point that they can afford, and have never been expected to justify their choice. So, in terms of enforcement, first and foremost we have made our position clear and shown that AI-assisted projects are not welcome on SourceHut. We will not be deploying any automated tools for detecting AI use or going out of our way to identify covert AI use in projects. One can lie or mislead us about their use of AI, but we want to signal to them that, if this behavior is discovered, they will lose the assumption of good faith and their account will be suspended. Therefore, it is not a wise choice to invest in SourceHut as the platform upon which this kind of software can be built, and we hope that reasonable users will make their choices accordingly. Existing projects on SourceHut which have made use of AI assistance in their work may remain on SourceHut provided that they adopt policies which are consistent with our terms of service for their future development. If you have used AI tools in your software before, and want to remain on SourceHut, there will be no retroactive enforcement, and we will take action against users who shame or harass those who have used AI in their work. Nonetheless we invite you to consider the arguments presented here and adjust the way you work in the future before you consider migrating away. If your project is affected Projects which utilize and intend to continue utilizing AI assistance after the policy change takes effect will be treated the same as most cases of ToS violation: with patience and understanding. If you or your project are affected by these changes and wish to continue making use of generative AI tools, you are required to migrate away from SourceHut at your earliest convenience. There is no deadline per-se, but you should have a viable migration plan in place as soon as possible. SourceHut is free and open-source software. You are welcome to follow the installation instructions https://man.sr.ht/installation.md to self-host it, or to find a community that operates their own instance and would welcome you there. We can also recommend Forgejo https://forgejo.org/ as an easier to deploy forge that may be suitable for your needs, and a list of public instances https://codeberg.org/forgejo-contrib/delightful-forgejo user-content-public-instances is available. If you prefer to retain the mailing list workflow, deploying a solution like GNU mailman may be suitable for your needs. Proprietary solutions like GitHub may also be viewed favorably by projects which support AI use, and provide many features to support AI users. Much of your account and project data can be exported from SourceHut and imported with minimal loss into other forges. You can transfer your data to another SourceHut instance with high fidelity, and other platforms should be able to import a subset of the data e.g. git repos to Forgejo or mailing list archives to GNU mailman . Please consult the hut https://sr.ht/~xenrox/hut/ tool for this purpose. If you require assistance in figuring out your migration plan, or want to make sure you have enough time to implement the changes while minimizing the disruption for your project, please don’t hesitate to ask us for help mailto:~sircmpwn/sr.ht-support@lists.sr.ht .