KDE is on the news because of a controversial proposal to define an official “AI” (LLM) policy (archived link). Other projects have tried their hand at similar policies and stances but, in my opinion, they miss the mark about the goal of such initiatives. I think that the point of these statements is shaping social norms and not micro-managing developer workflows. They should be about signalling what kind of behavior we want, and what kind we reject.
This proposal does not go into detail about the many problems that LLM have caused to society, workers, the environment. It goes without saying that all these ills are fundamentally opposed to the humanist spirit of GNOME.
With all that in mind, here is what I personally think a GNOME LLM policy could be:
A GNOME Project LLM Policy
The GNOME Project prioritizes the social and human aspects of
collective software creation. Therefore:
1. LLMs ("AI") can not be used to create or modify
anything submitted to GNOME, or hosted on GNOME infrastructure.
You might be asked to prove your code meets this requirement.
You might be banned for trying to circumvent this policy.Example Guidelines for Contributors
These are just a draft of the kind of criteria one could use to evaluate if a submission fits the policy.
- You must be able to personally reason and explain your changes
- You must be able to demonstrate knowledge of the problem space you are working on
- You must solve the underlying issue, not just its symptoms
- You must respect the time of fellow contributors
- You must not impersonate yourself through chatbots, agents, or other automated systems
This Is About The Future Of GNOME
GNOME is not just software that happens to ship every six months. That is just a delusion we have been holding up for the last 30 years to keep our loose group of colleagues, friends, and acquaintances, together.
GNOME exists as a collective that find joy in reaching beyond our individual limitations to achieve something bigger. These people, this joy, are the whole point of the project. Contributors are not payroll, a liability, that we hope to downsize next quarter.
“Come do free labor for a handful of corporations by reviewing chatbot output in your free time” is not an attractive proposition to young talented people in 2026. If we want GNOME to continue we need to create an attractive and inviting social space where people are valued as people.
Just like the Foundation is moving to individual donations to stop depending on just a handful of companies, we need to look for the next 100 people that will donate a tiny bit of their time, instead of hoping that corporations will keep 10 overworked engineers on staff. We already have seen how companies will happily abandon a whole chunk of GNOME on a whim.
GNOME is not just software, and it should protect the social and human aspects that make it special. Our success metric is the community and social bonds we create. In the most literal sense GNOME is about the journey and the friends we make along the way.
Free Palestine.
FAQ
How do you enforce this?
You can not. People will still send LLM generated code. This policy makes it explicit that we do not welcome these careless submissions. We have a Code of Conduct that is 80% about telling other people what our values are, and 20% about handling unwanted behavior (“enforcing”). This is similar.
What if people simply lie about not using LLMs?
This is the same problem as authorship, in the copyright sense. Whenever we receive new code we have to assume that “beyond a reasonable doubt” said code has been authored by the person contributing it. We make our best guess. The attached guidelines are a suggestion to make these new guesses.
Ok. But what if people are really good at lying?
This policy is about the majority that will not even try to lie. See previous questions.
fwiw, nearly all work I have done for the last year or so and nearly all work I will do going forward for the rest of my career as far as I can predict at this point, will heavily leverage LLMs. From review, to pr submission, to code generation, to commit messages, everything I do and will do going forward barring circumstances I can’t for see will lean heavily on LLMs.
Not just coding, but general system usage as well. I don’t type “find and sed commands anymore, I ask codex. That doesn’t mean I can’t stand behind work with my name on it, it doesn’t mean i’m putting stuff out there I don’t understand. It doesn’t mean i’m tricking people or wasting anyone’s time. At the end of the day what matters is that the output matches or exceeds the quality that was there before LLMs became mainstream.
The truth is, as far as I can tell, AI is not going away, and the number of people who use them will asymptotically approach 100% within a few years at most. A sea change is taking over the world and we need to embrace it for the value it provides, and try our best to deflect the problems it creates.
So while I whole heartedly agree that we shouldn’t waste anyone’s time putting output out there that doesn’t meet a bar, policy restricting the provenance of that sort of output cannot work and isn’t even a good idea. I agree with all your bullet points but the last one.
Another adjacent point you didn’t bring up directly, but I want to mention from my soap box is “disclosure” in commit messages or MR texts. I have no desire to fill commit histories with free advertising for AI companies, and code generation tools are just tools, not authors, even if they are generating much or most of the code I end up using from my prompt texts.
The buck stops at me with my contributions, and just as I won’t agree to use emacs instead of vim to write code, even if project policy requires it, I can’t see myself agreeing not to use code generation tools of my choice, either, and if it’s codex, or grok or claude code, co-pilot or cursor, or vim-enhanced or neovim or VSCode or some mixture of all of the above, that’s really between me and my screen, no one else.
“I have no desire to fill commit histories with free advertising for AI companies”
Is it not possible to disclose an AI was used without mentioning the specific one?
“code generation tools are just tools”
https://deadsimpletech.com/blog/no-such-thing-as-just-a-tool
Hi Ray, thanks for the lengthy response.
As I said in the opening paragraph of post, I don’t think these policies should be about the developer’s work flow, nor is it possible or should be policies how you made a patch.
This is entirely about how people behave online and our common spaces not be littered with diarrhea and slop all over the place.
I appreciate that you review and take ownership of your changes, but that’s not the status quo. It is also the bare minimum of what we all expect on our community, and have had expected before these things existed.
I won’t go into detail as its much bigger topic, but the reason I proposed a ban is cause anything else is a permissive policy that only results into normalizing ChatGPT and Claude and invites slop. The moment you say “If you use these things, follow X and Y” you make it acceptable to use them. I don’t think that should be something that the community should be promoting and accepting.
Related: https://tante.cc/2026/06/30/trying-to-manufacture-permission/
No tool exists in isolation of its environment or is neutral. There are so many issues with them that’s its hard to keep a list of them all, but I think you’d have a hard time denying that Claude and ChatGPT are incredibly biased, racist, addictive, actively kill people, destroying every single aspect of society and on the lower scale of issues, they make our lives as Free Software maintainers plain miserable.
I can’t stop anyone from using them nor care to judge them, much like I don’t care if they are smoking cigarettes when making a patch. The only thing I ask is that they don’t blow the smoke on my face. I think that’s the bare minimum.
I agree with you. I think AI should be banned from any project that places any importance whatsoever on human interaction. I think it’s irrelevant whether AI produces slop or not. If AI produced perfect code every time, I would still want to ban it, because:
1. AI encourages hyper-independence and less human interaction. Even though tools make individuals more independent, there’s a point of diminishing returns and the AI that writes perfect code means no one has to rely on anyone else. Of course, AI can’t do that yet but what’s the point of doing anything in life if the end result is just individuals using software in isolation without any interaction.
2. Part of the point of software is to use it to do something useful. That’s true. But part of it is also to build a community behind it. I think we shouldn’t lose sight of that, but AI goes against that in spirit and for that reason it should also be banned.
3. AI itself represents the automation of tasks that at one time required creative human output. By supporting developers who use AI and accepting AI, you are also destroying spaces and opportunities for others to express their own uniqueness.
Therefore, the very spirit of AI goes against a healthy human society and a healthy place to come together in our common love for creating things. Even if it has technical benefits in the short term such as finding bugs, making things more secure, allowing the drudgery to be lifted from the lives of senior programmers, NONE of that is worth the negative consequences of AI that is embodied in its very core.
Many distributions and desktops embracing AI-generated submissions with open arms has caused me to be quite alienated from even affiliating with these projects anymore. I’ve begun actively seeking distributions and desktops that lean towards the negative side of AI, I just don’t really want anything to do with it.
I don’t understand why it’s being pushed as something that is inevitable, we have coded for decades before it just fine. “Eternal acceleration” and min-maxing code isn’t necessary and is quite detrimental to humankind in my opinion as a decade-long Linux user.
As someone who’s not a maintainer or developer, my voice doesn’t hold any sway for the most part, but recent events in Linux spaces have me seriously considering learning to code to put my money where my mouth is and work towards keeping humans the centerpiece of all of this.