So I accidentally triggered an online shitstorm in the process of trying to craft a set of more restrictive LLM usage guidelines for KDE. Sorry about that. 
I’m going to leave the comments on this post open, but I will request that everyone who wants to comment read and understand the entire post first.
So anyway…
What happened?
Here’s a verifiable timeline of events:
- Several weeks ago, a KDE developer started a mailing list thread asking about how we should handle the flood of AI slop merge requests.
- Everyone agreed it was a problem, and that we don’t want that crap.
- There was less certainty about the best way to handle it socially.
- Several wished for an official policy so we didn’t all have to figure it out on our own.
- I proposed a set of draft guidelines making it quite clear that slop merge requests are unwanted, and can be ignored or closed.
- It received a generally positive reception.
- Another KDE developer requested that we workshop it somewhere more modern and capable.
- I got busy and didn’t do so.
- Several weeks later, various people on the internet discovered that a talk proposing “a lovable, sovereign, AI-native KDE” had been accepted at Akademy 2026 and created drama around it.
- Akademy happened, and the talk was given to what I’m told was a fairly chilly reception.
- The next day, a workshop was held about the topic, also receiving a chilly reception.
- Realizing this topic was not going away, later that day I re-opened a second draft of my LLM guidelines on invent.kde.org.
- KDE contributors began to discuss the draft proposal.
- Two people unknown to any KDE contributors appeared and began fighting with one another about the broader topic of the morality of AI, not the proposed guidelines.
- One of those people doxxed the other.
- Comments on the draft proposal were locked to only KDE developers.
- The person who doxxed the other was banned.
- Later, the identity of the doxxed person was verified, and they were also banned for their conduct outside of KDE being so unacceptable that people inside KDE did not want to associate with them.
- Someone else outside of KDE set up https://kdeforpeople.com in an attempt to brigade KDE into banning LLMs.
- A bunch of people signed onto it, almost none of whom are known KDE contributors (I see one person I know in there).
- The topic was picked up on social media and the press with… varying levels of accuracy.
- The draft proposal was removed and the whole topic hidden.
TL;DR:
A bunch of people mostly outside of KDE who disapprove of LLM usage derailed KDE’s attempt to add restrictions to LLM usage.
Yep, that’s where we’re at in the state of online discourse around AI.
I’m sorry
After everything I wrote last month about keeping your community safe and containing damage so it doesn’t spill out into public view, I feel pretty bad that my attempt to craft a compromise policy around LLM usage in KDE (which would have tightened it!) backfired and created a storm of drama.
It wasn’t my intention, and I apologize for my role in it. If I make another attempt to craft a policy that provides certainty and cover for maintainers who want to reject AI slop, I will be even more careful to avoid creating unnecessary drama. If someone else wants to, I will try my best to help them with the same.
More facts
People jumped to some pretty wild conclusions in the various social media threads about the topic that I saw. Before we continue, let me set some more records straight:
- KDE e.V. has not taken money from DHH or Omarchy. I also have not, either personally, or in my business.
- KDE e.V. has taken money from Framework, but only after verifying that Framework had severed their (already very limited) ties with DHH and Omarchy. Also, all of this happened a year ago, before Omarchy received a ludicrous amount of money from a veritable rogues gallery of villains.
- Today, Omarchy is radioactive in KDE and adjacent communities I’m aware of. Nobody I know wants to have anything to do with them.
- The “lovable, sovereign, AI-native KDE” idea was presented by two people important to KDE in years past, but with no contributions in over a decade. Their idea does not reflect the overall direction of KDE or Plasma, and I don’t think it ever will. If “a lovable, sovereign, AI-native KDE” freaks you out, I believe it is completely reasonable and safe to ignore.
What on earth is going on here?
I know people are really riled up about AI. I get it. The negative effects of LLM usage seem to become more obvious by the day:
- They lower the barrier to entry for people who have no idea what they’re doing to DDoS FOSS maintainers with a tidal wave of crap. And this was the actual reason for the proposed restrictions.
- They make it easy for lazy people to avoid learning communication and job skills.
- They make it easy for lonely people to form unhealthy para-social relationships.
- The companies behind the big ones are really bad.
- Some of their training data was stolen.
- Some of their training data that was not stolen results in the destruction of rare books.
- The process of training them on freely available online data imposes extra hosting costs and slows down access for humans, acting as economic free riders.
- The license of source code they produce is legally unclear.
- Their resource usage frequently strains the infrastructure of the communities their data centers are located in or near.
- Their data centers are using enough electricity to blunt and reverse prior positive global trends of lower electricity demand and greener electricity generation.
- The specialized hardware they use has monopolized global production lines, massively inflating the price of normal consumer and business computer hardware.
- The enormous amount of capital allocated to them may be impossible to recoup, with the potential to eventually cause a worldwide economic crash.
- They are either mandated or voluntarily used by so many people already, that many who do not use them fear they will “fall behind” or “become obsolete” despite concerns with any of the above.
So yeah, bad stuff. I completely understand why a lot of people have problems with LLMs. I have these concerns as well.
I also understand why a lot of experts in their field really like LLMs. These are very powerful and sophisticated tools. Capable people with decades of experience can successfully automate away a lot of tedious work, and still end up with a final product identical to what they would have done by hand.
I don’t use LLMs for automating work myself, but I do understand the appeal to many experts.
And I also understand that LLMs can be responsibly used in really useful ways for general tasks like learning a nebulous skill, retrieving deeply-buried information you can only describe in unspecific ways, or helping you troubleshoot a problem with unclear steps or parameters.
At the same time, I think we have to acknowledge that irresponsible LLM usage by people who are not yet experts is a disaster for their learning, expertise, social skills, and emotional health.
In other words, this is a hard problem.
What on earth do we do?
How are we expected to handle a technology that is:
- Very powerful
- Genuinely useful for various things
- Dangerous, de-skilling, and/or addictive in the wrong hands
- Full of really bad externalities
- Broadly available worldwide at effectively zero cost
It’s like 100 billion toaster-sized nuclear reactors dropped from the sky one day. The very least you can say is that this would be very de-stabilizing!
The world is changing fast and I have no idea what’s going to happen. But I don’t think the genie is going to be put back in the bottle. I don’t know what to do, and that’s scary. And I have a lot of empathy for anyone else who’s scared by this, too. We’re in this boat together.
What can you do to help?
If I or someone else tries again to propose a set of usage guidelines that restrict inexpert usage of LLMs, with the goal that KDE receives fewer of the AI slop merge requests that are burning people out, please don’t make things worse.
Please don’t derail the process.
Please don’t take it as an opportunity to fight over the broader topic of AI in general.
Please don’t try to burn each other to the ground over it.
Please don’t doxx or send hate mail to anyone over it.
Please, just be kind. We’re all trying our best over here. No one in FOSS is a cackling techbro… outside of maybe Omarchy. 😬 The rest of us are trying to figure out what to do about the nuclear reactor that fell into our living room without killing ourselves, hurting anyone else, or becoming unemployable.
If at this point you want to post a comment, please be kind and treat the topic with an adult level of maturity. I believe in you. 🙂
Well explained overview. I hope given a small break to let things settle , that a group would pick this back up in a productive way so that guidelines are created to protect original work.
❤️🙌
LikeLiked by 1 person
It’s frustrating how you totally discount the desires of KDE users. Are you just developing KDE for yourself? Or do you think users don’t care how you develop it?
LikeLike
For the benefit of education, this is an example of a comment that’s not treating the topic with an adult level of maturity.
I’m keeping it here for illustrative purposes, but all additional ones will be removed.
Reminder to behave like kind adults. 💙
LikeLike
Well, everyone has some kind of opinion on this topic, but nobody wants to be in your shoes right now 😉
But at the end of the day, it’s up to the KDE contributors to choose what is the best tool (AI or not) for them. We all have trust them for preserving KDE as Kool as it actually is.
So maybe they should vote for the proposition they prefer, like Debian did. And maybe its impossible to check if someone is using AI for anything. Good luck Nate anyway.
LikeLiked by 1 person
Yeah, anything that fails to achieve consensus (KDE’s typical standard) will probably end up put to a vote. It can be a reasonable path forward, especially if the result is super lopsided, like 85-15 or something. IMO something like 55-45 would be terrible though, and would indicate that the proposal isn’t good enough. In that case, I’d retract it and try again.
LikeLike
Thanks for your article Nate.
I trust both you and KDE in your mission and I never doubted you 😉
You can count on my support (as I am not a programmer, it will be mostly financial, but still you all deserve it! ;))
LikeLiked by 1 person
I guess that the speeches by Mr. Sanders and Mr. Bannon at the invitation of Mr. Tegmark lately escalated the public debate and hysteria. Though I think we should have a nuanced examination of the issue. AI can help humans save lots of ressources, not only lifetime but also re-compilation and debugging time. Also one should not compare or confuse the natural resource consumption of AI as coding assistance or simple knowledge work with the resource consumption of endless slop videos to provide liberal democracies with entertainment, gaming, product ads or political disinformation campaigns, just to name a few.
Also, since I practice Software Development for around 30 years now, my mind is literally code when I reason about it. But sometimes I just do not want to write prosa to explain it in human transferable language. So LLMs can help to formalize concepts in a broadly conceivable and compact way. But please, configure your agents to formalize short and sweet, to use self-explanatory symbol names and only comment code if needed. I neither have no time to read AI poetry, nor do I have the display size to do so in a comfortable manner. We should encourage people to use AI as a tool for enlightenment (in the spirit of German Idealism) and to grow with it, but to also keep in mind that life is short, the craft so long to learn, and that cooperation and collaboration should hopefully imply the intention to make each others life longer and easier.
LikeLike
PS: what about an AGENTS.md file template for KDE projects to declare coding style and other values, so that project maintainers can adapt it if they wish?
LikeLike
I don’t use this technology for programming myself (I hardly get in any programming time at all anymore, sadly), so I’m not really the right person to talk to about that. I think the KDE developers who actually do most of the work should be the ones to decide anything like that.
LikeLike
Thank you Nate, I really appreciate the explanation. I will go fully on record and say that I am absolutely, vehemently against LLMs and AI usage in software projects. Anyway, I do want to touch on this point:
“A bunch of people signed onto it, almost none of whom are known KDE contributors.”
I think that the opinions of KDE users should matter more than it seems they do, purely judging by this wording. That is also one of the ideas behind “KDE for People”, right? People as well, not just devs. These are people who care about the quality of KDE, the planet, and the considerable overlap between LLMs and fascism in tech. Just to name a few points.
Perhaps consider why the people using KDE are going out of their way to voice their frustration against LLM usage in KDE, despite not being developers themselves.
Enabling people to submit vibe-coded patches they can’t make heads or tails of is one issue, but not the only one. There are other concerns too, one example being the environmental impact of LLM/”AI” usage. KDE Eco already voiced their concern about it.
I also theorise that some of the outrage might have been sparked by how the KDE social media account is handling things on Mastodon. The person (people?) in charge of it might not be wording themselves as well as they should…
But all in all Nate, I think you have a very level-headed mindset and glad that you continue to be a voice of reason within KDE. Again, I appreciate that. Whatever the outcome I also agree KDE as a whole should take their time to brainstorm a solution to the issue.
LikeLike
Every KDE user should matter, but without developers no kde.
Wouldn’t it be better define where and how they allowed to use these technics, instead its current state?
Its like everyone use cars (that are obvious bad for the planet) and the kde developers forced to use horses.
LikeLike
You’re welcome! And to be clear, I think what users want is definitely very important!
But what percentage of KDE users want to ban all AI usage? 75%? 30%? 5%? 0.1%? We actually have no way of knowing, and each of those hypothetical levels of support would imply a very different course of action if they were the real one.
Our flawed human brains give a a minority of noise more weight than a majority of silence, which is why pressure campaigns often work. But that doesn’t mean this is a good method of decision-making.
Users need to understand that the relationship they have with FOSS developers is very different than the one they have with Apple, Amazon, Walmart, etc. KDE software is literally given away for free; it’s a gift. And the social contract for receiving a gift is that you show appreciation for it and for the person who gave it to you. This applies even if you don’t like it, or if you do but think it’s imperfect or could have been made better, or sourced more ethically, or whatever. Doesn’t matter! You don’t “look a gift horse in the mouth”.
Because if, instead of showing appreciation for the gift, you complain about any of those things or the people who made it and gave it to you, you’re violating the social contract.
I’m not saying never complain or make suggestions or passionately advocate for change. I’m saying understand that when you do so for a FOSS project you are not contributing anything to, you’re violating the social contract for receiving a gift, so your perspective had better be really well-considered, reasonable, and actionable.
And if you don’t get your way? That’s life, sometimes it happens. Ultimately it’s up to the people who actually do the work to filter their perception of user desires through other lenses as well: what’s possible, what’s feasible, what makes sense for the project in their estimation, what there’s time to do, and so on. Without them, there is no KDE to argue over! So we should be really careful about burning them out, whether that’s with AI slop or opportunistic pressure campaigns to convince them to ban all AI.
LikeLike
I feel it’s a bit silly to paint kdeforpeople as “brigading”, it’s essentially an online petition/open letter.
It’s also a bit silly and kind of a bad look that when someone brought up their concerns that the voices of the people outside of the core development team aren’t taken into consideration (which I know isn’t true because I’ve been following this whole thing from the start, but other people are likely unaware, or distrustful considering everything going on right now in the world) you highlighted said concerns by just saying they’re not being mature about it.
I get that it’s a stressful time, it’s been a stressful week for basically everyone, but I don’t think that was the right way to handle it.
People are worried. People don’t want KDE to become yet another vibe-coded project, people love KDE and these concerns are proof of that in themselves.
LikeLike
Your stance is clear Nat, and I appreciate that, but I believe we should always reflect on what makes our community unique.
As things really do start with people we don’t know, they spread like wildfire through dry straw; no one can control the fire, but everyone wants to have their say on… fire safety measures, theories of fire, fluid dynamics and the temperatures of dry straw.
All of this has ended up creating our very own KDE El Niño.Perhaps, just perhaps, if everything had been more transparent… we’d have had a different outcome. We mustn’t be afraid to lay everything out in the open; the community only grows through genuine dialogue.Anyway, let’s move on – lesson learnt?
LikeLike
I’m glad KDE is taking a “human first” approach to banning slop.
I recently unfortunately learned that XFCE had taken up using LLM’s themselves. Having used XFCE ever since GNOME 3 released I was rather disheartened.
I think in future I’ll be sticking to KDE for my Linux adventures
LikeLike
Nate, you hopefully know that the Invent kerfuffle was not your fault, it’s unfortunately pretty common for commenters to get fighty on the internet. I’m glad to see KDE devs taking it slow and deciding what works best for y’all.
LikeLike
A major thing that makes me strongly disagree with that guideline was kinda basically the modus operandi being “dont tell us about AI usage”, by removing the commonly established and by other projects used way for that disclosure, with that assisted-by tag or even any other way, because “its advertising”
The looks also are “we are closing the discussion behind closed doors, and only to developers, when KDE actively pulls in the work of artists and other community parts, and the look is that it disregards its users that didnt feel compelled to comment on other issues.
The active look of events is like “attempt to pull a less than popular policy while everyone else is busy because they are at Akademy, or busy with other 30th anniversary things”.
Pulling up ones pseudonymous twitter is not doxxing.
LikeLike
I really appreciate the thought and care you put into this response. Thank you. I hope you’re representative of KDE developers as a whole!
LikeLiked by 1 person
I’d like to reiterate what Chip above said. I think there is a disconnect between devs and users. Why not put up a poll and ask everyone and see…the potential is alienating a bunch of users and pushing us to other WM and DE’s and in the process fracturing trust with I’d like to think a fairly loyal user base. The Linus approach has pissed off a bunch of people, and we are seeing slow movement away from AI…to bsd, etc and I have no doubt another iteration will come as an alternative.
I don’t think KDE wants or should tell its users to “fork it”
LikeLike