KDE and AI, and you, and me

FOSS culture, Inside KDE 6 Minutes

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:

  1. Several weeks ago, a KDE developer started a mailing list thread asking about how we should handle the flood of AI slop merge requests.
  2. Everyone agreed it was a problem, and that we don’t want that crap.
  3. There was less certainty about the best way to handle it socially.
  4. Several wished for an official policy so we didn’t all have to figure it out on our own.
  5. I proposed a set of draft guidelines making it quite clear that slop merge requests are unwanted, and can be ignored or closed.
  6. It received a generally positive reception.
  7. Another KDE developer requested that we workshop it somewhere more modern and capable.
  8. I got busy and didn’t do so.
  9. 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.
  10. Akademy happened, and the talk was given to what I’m told was a fairly chilly reception.
  11. The next day, a workshop was held about the topic, also receiving a chilly reception.
  12. Realizing this topic was not going away, later that day I re-opened a second draft of my LLM guidelines on invent.kde.org.
  13. KDE contributors began to discuss the draft proposal.
  14. 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.
  15. One of those people doxxed the other (edit: I have been informed that this is not really what doxxing is) exposed the other person’s odious Twitter posts in an effort to discredit them.
  16. Comments on the draft proposal were locked to only KDE developers — one of two tools that that Gitlab provides for this kind of thing, the other being to hide it entirely.
  17. The person who doxxed exposed the other was banned.
  18. Later, the identity of the doxxed exposed 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.
  19. Someone else outside of KDE set up https://kdeforpeople.com in an attempt to brigade (edit: I have been convinced that this was too strong a word to use here) pressure KDE into banning LLMs.
  20. A bunch of people signed onto it, almost none of whom are known KDE contributors (I see one person I know in there).
  21. The topic was picked up on social media and the press with… varying levels of accuracy.
  22. 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 decades past, but who had not made any contributions recently besides this Akademy talk. 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.
  • They have been discovered autonomously hacking rival companies.
  • 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:

  1. Very powerful
  2. Genuinely useful for various things
  3. Dangerous, de-skilling, and/or addictive in the wrong hands
  4. Full of really bad externalities
  5. 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 actually doxx or send hate mail to anyone over it, or try to get them fired from their job.

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. 🙂

Unknown's avatar

Published by Nate

Published

109 thoughts on “KDE and AI, and you, and me”

    1. English translation:

      enough with your fashtech. Ban AI in KDE or we will kill you and your children.

      I’m leaving this comment up too, as an example of where an overheated discussion can take people.

      Is this really what anyone wants? Does anyone think this is convincing?

      I understand it’s a charged topic, but hopefully everyone can agree this is a really bad place to go. Maybe it’s time to turn down the heat and walk a few steps back from the cliff.

      Like

    2. We get that you’re angry, but don’t you feel that death threats are a bit exaggerated?

      Kindly,

      Mark.

      Auf wiedersehen

      Like

  2. I am a KDE user, private linux distribution maintainer, and a system administrator for multiple other KDE users.

    My users and I implicitly have a stake in KDE, its future, and how it is developed. None of us are interested in being KDE contributors, particularly after this incident.

    If my users and I have opinions about topics like this and want to express them, and issues like this or “KDE For People” (which I have users who have signed onto) are not an appropriate avenue to express what we want to see, then can you tell us what *would* have been appropriate?

    Right now, this still reads to me like “we aren’t interested in your feedback, KDE is for us developers who have earned the privilege of caring about how their system works, we already know everything, please stop yelling at us”. This may not be how you intended to be heard, but it is what you are saying, and it is hard for some not to feel insulted.

    Like

    Reply

  3. One thing we can take out of it is that nobody in the KDE community wants AI Slop. There’s obviously a disagreement on how to approach that and how far.

    Moving forward it would make sense to keep these discussions in the open/transparent but ask for specific audiences on gitlab and having a public forum for users (polls, or discussion forums) to get that feedback/opinion. It seems like there is a vocal group that feels any AI use is morally wrong and giving them that opportunity to express that as a user base will lead to better discourse from contributor channels that are and should remain transparent.

    On the subject of transparency, I disagree that having a disclaimer is advertisement. It helps to further limit contributions and analysis if a commit leads to a vulnerability or breaking code. It could also lead to evidence/data if current llm-supported tools are statistically pushing vulnerable or poor quality code (or the opposite). That information could then be used to support future policies.

    Like

    Reply

  4. I respect that you are trying to curb AI usage. I think you and all the others who are doing it in this manner are trying to perform damage mitigation in a “we do not negotiate with terrorists” situation, and are not treating it with the actual concern warranted. Adopt the LLM policy outlined here: https://blogs.gnome.org/alatiera/2026/09/23/the-gnome-llm-policy-that-i-want/. This isn’t difficult. These techologies are only “useful” in a rate race way. More and more knowledgable coders are admitting that the cost of maintaining even “good” LLM generated code outweighs the time savings. Unfortunately there is probably no way of disentangling them from security patching due to the fact that bad actors will forever onward have the open source models that already exist and therefore blue forces will need to check their work against those models for holes, but that is a horrid state of affairs that every lover of software and every company using LLMs should be shamed for allowing to exist. Starve the tools. Make it clear that they are not tolerated, regardless of one’s experience level. Let. them. die.

    Additionally, I find it rather offensive that you discount the existence of the website listing individuals who wish for no LLM usage in KDE. I understand you were talking to other developers, but ultimately it’s your users who have responded to you, and while there are times where the users may sincerely not know best it still is extremely rude to act like their voice doesn’t matter when they honestly will be spending thousands, if not millions, more time interacting with the KDE products than the developers do in aggrigate.

    Like

    Reply

    1. This isn’t difficult.

      That is merely a shorthand for “it is actually difficult, but I refuse to acknowledge that.”

      Things indeed get easy when you just invalidate any other stances, but to effectively do that, you need overwhelming force and you need that constantly from now on until the end of time.

      So not doing that is the only sensible option if one wants to actually change minds.

      Like

  5. I think the approach was fine. It is just a very contentious issue spreading throughout the Linux and FOSS space right now. Mix in some that wanted to derail it from the start and you get the mess we had.

    As a Linux distro maintainer, we support the efforts that KDE is making to find a balance, and the approach, in my opinion, was balanced. Some want to only accept extremes, regardless of what others on the team and community want. That never works. Finding a balance approach will certainly still cause some to be upset or disappointed, but that is how it works, because it is for the people, on both sides of this.

    Like

    Reply

  6. Hi Nate, I appreciate that you are trying to defuse tension, but I think you are missing a big part of why your post isn’t doing that for lots of people, but rather reinforcing their fears. Plenty of other commenters have already discussed the original inflammatory language of “doxxing” and “brigading”, as well as the dismissive attitude toward users respectfully expressing their concerns. I won’t dwell on these, as I think they have been discussed enough, and I appreciate your modifications to your original post.

    In response to an earlier comment, you mentioned you don’t have time for being on the fediverse. Honestly, that’s great that you don’t have time for it and are putting so much time and love into KDE! But I also think that’s part of why you’re missing a big part of the context.

    We’ve been watching, in project after project, people in FOSS that we thought that we could trust give in and allow slop. Linux, Debian, LLVM, Rust, Python, Blender … and now, KDE? You say that your proposal was about restricting LLM use, but I don’t think you understand why it’s being received the opposite way. Many people understand LLM use in binary all-or-nothing terms. It is extremely frustrating and demoralizing to witness so many people fail to take a simple, unambiguous stand against something that, to many of us, seems so utterly obvious how the cons vastly outweigh the pros. Some of that is motivated by moral purity testing. But it’s not just that. It’s because of the experience of watching what is happening all across the FOSS world, and the wider world.

    Crucially, many LLM opponents don’t view it binary terms because they want to, but because of the actual lived experience of the behavior of LLM users. Give them a millimeter, and they take a lightyear. Already in the comments on this post there was someone advocating for adding AGENTS.md files to KDE repositories. The presence of AGENTS.md files is interpreted by both proponents and detractors of LLMs as a clear sign that LLM usage is not only allowed, but encouraged. It is hard to argue against merging a pull request adding an AGENTS.md files without a complete, uncompromising ban on *all* LLM usage. The policy you proposed would not do that; it would leave the door open to things like AGENTS.md files.

    Once those are there, it is like a magnet attracting more LLM users, and the slop PRs multiply. Despite the policy ostensibly put in place to restrict LLM usage, people who don’t want to deal with LLMs get more burnt out and some leave. The more that leave, the more LLM users come, and before you know it, the community gets taken over by LLM proponents. I strongly encourage you to read this blog post detailing how this process is playing out in the Bevy game engine community: https://blog.fallible.net/it-breaks-a-village/ . I would also encourage looking at what is going on in the Linux kernel community: https://floss.social/@pinchartl/117320477532952626

    You say that there is not consensus for a complete ban on all LLM usage. But there is a flip side to that: is there a consensus for allowing any LLM usage at all? How can you know, when in the discussion of your proposal, anyone advocating for a stricter ban was condemned as “off topic”? Can there be a consensus on this topic at all in such a large group of people? I doubt it. I think some fracturing of the community is inevitable. The question facing KDE is: which way will it fracture? Will the apparently small minority of contributors that are using LLMs today be asked to stop doing that in KDE spaces? Would they comply, would they leave, or would they lie about continuing to use LLMs? It doesn’t seem that has been discussed seriously. Or will everyone else be told, just like everywhere else in life, that they have to tolerate LLM usage in the shared space? Already in the comments on this blog at least one contributor has voiced that they will leave if LLM usage is tolerated.

    Thanks for engaging and keeping the comments open here. I’ve been using KDE for 20 years. I’ve made a few small contributions of code and bug reports over the years, and I’ve contributed to upstream code that KDE uses. I hope that unlike so many other projects, KDE can take a simple, unambiguous stand on this, and I can unambiguously feel good about continuing to use KDE. I think the best next step would be a serious conversation considering a complete ban on all LLM usage in KDE, and see what the response is to that compared to presuming a compromise is the best starting point. If it’s true that most KDE contributors do not want to deal with slop, I presume only a few would have a problem with a complete ban. If you are looking for an example of such a policy, the best I have seen is Twisted’s policy: https://docs.twisted.org/en/latest/development/ai-policy.html

    Like

    Reply

  7. I’m late to the drama and would preface by saying: I’m not a contributor. Maybe some bug reports here and there.

    All I wanted to say that I still trust KDE and that proposal (2nd draft I read I think) was the most “middle-of-the-road-this-is-the-baseline-most-of-us-agree-on” path (in a good way IMO) and was pretty good for an initial draft. It was natural to have people with stronger stances on things like the pros and cons of using tech like LLMs and it’s impacts and that’s fine. But some of the responses in that post and on other areas like mastodon (not surprised on this one) were very staunch and …. IDK almost tribal. There’s the usual “oh there seems to be a third degree connection to company x therefore your organization must be <insert some connection to some bad stuff>” or “organisation has fallen off/compromised/whatever”. Like, at least read and understand things before replying?

    Anyway, I completely see valid points on either direction. It’s valid (to me) to say LLMs waste too much resources and generally result in more chaos, more nuisance and slop, unless used by capable hands. I’m personally tired of just seeing people fight and burn out of filtering AI slop in their projects. Not to mention the output being in a legal grey area, and the source data essentially being stolen.

    It’s also valid (IMO) to see it in the context of rest of the things we do which promote similar kind of damage like using planes/cars/plastic/non-renewable energy etc. and going, “why only target LLM usage? We don’t get this tribal over those other things, even if we probably should. But until we do, it’s not fair to single out AI.”

    I’m sure there is an amicable path to take here. KDE isn’t a singular entity, even when presented as one. Just like a varied userbase, the contributorbase is also diverse and all have their own lines in the sand regarding what is acceptable and what direction should projects under the KDE umbrella go. And personally I trust KDE folks to sort it out.

    P.S. It’s probably better that you mostly stay away from mastodon. Lot’s of good people, don’t get me wrong. But sometimes it feels like mastodon is the reddit for FOSS :-P. Okay won’t say much or I’ll be nuked if they find me hehe.

    Take care team, and remember that there are a lot of happy users who aren’t here, it’s not all negative.

    Like

    Reply

  8. You state that people were talking about “the broader topic of the morality of AI, not the proposed guidelines”, but the morality of AI is a central part of creating and assessing an AI policy. I understand that you originally proposed the policy for practical workflow reasons, but that’s not the only impact of the policy. Talking about AI morals isn’t getting sidetracked. It’s getting to the heart of the policy. I think that’s the fundamental disconnect between the KDE team and the KDE users. That’s why locking down the discussion to developers was seen as censorship.

    And the fact that you describe kdeforpeople as an attempt to “brigade”/”pressure” KDE compounds that. You mention that you only recognize one person in the signators. That’s the point. The point is to elevate the voices of people who have a legitimate interest in the quality and security of KDE, but who are not part of your in-group.

    Those two facts can reasonably be summarized as “discount[ing] the desires of KDE users”. The expected emotional response to that is “frustrati[on]”. And it also raises the question of what the point of KDE even is. I’ve always believed that open source software should be for its users. Open source is a gift to the users; they’re not entitled to it. But a gift given without consideration to the recipient is still a bad thing, actually. Condensed into pithy rhetorical questions: “Are you just developing KDE for yourself? Or do you think users don’t care how you develop it?”

    It’s no mistake that those summarizing quotes put together are the totality of a comment you described as “not treating the topic with an adult level of maturity”. How much are we expected to expound on the core issue to be viewed as legitimate? How much effort do you expect people to put into elaboration, and how does that compare to the effort you put into trying to understand why people say what they say? Are you setting a fair standard for this discussion?

    An aside: you also referred to open source as a “gift” in one of your responses, and you emphasized the social contract of gifts. But software differs from physical gifts in a crucial way: they, by necessity (security updates), need to change after they’ve been given. Once you gift something to someone, it becomes theirs, and if you changed a physical gift in a way the new owner doesn’t want, they could haul you to court for vandalism or conversion. Software isn’t held to such a strong standard and likely shouldn’t be, but it should still be maintained with the interests of the people you gifted it to. And even discounting all that: your description of the social contract of gifts, as popular as it is, is something I disagree with. You will rarely find an idiom/proverb that I agree with. Every gift also comes with a burden, even if it is nothing more than the space it takes up on a shelf. In the case of AI in open source software, the burdens include security concerns, copyright concerns, and moral concerns. Treating people as ungrateful for caring about the burden you’re trying to place on them is not a virtuous stance.

    Lastly, when making a proposal (in KDE or otherwise), a clear statement of intent is wise. Clearly stating the intent was to empower maintainers to reject slop would have curbed a lot of the backlash, since a lot of users will view “no LLM usage” as the default. However, there still would’ve been criticism of the policy not taking a hardline stance, since, frankly, a hardline stance against AI is a very reasonable stance that a lot of users are going to expect.

    Like

    Reply

  9. Thank you for the blog post, but it still leaves a weird feeling with me. Two angles I think are missing here:

    • KDE software is incredibly important to people. Plasma is one of the software that makes my computer useful – and my computer is important to my everyday life. I’m depending on you (meaning the KDE devs) to not screw that up by eg. merging untested slop code or adding weird AI-features to my desktop. I now understood that you are not planning on doing this.
    • The dismissiveness in the GitLab thread and on Mastodon have made me question whether I have a say in this. I’ve been using Plasma for about 15 years now. I’ve reported some issues in that time. I’ve recommended Plasma-based Linux distributions to pretty much everyone I know. I’ve organized an End of 10 event last year. Am I part of the “KDE community”? Or do I have to write code to actually matter?

    Witnessing all of this produced a feeling of helplessness and made me wonder whether I need to cancel my recurring donations and look for other desktops, thankfully I currently don’t seem to need to do this.

    I don’t have any real problem with the actual policy proposal, but I thought it was a bit too permissive on LLM usage.

    Like

    Reply

  11. This isn’t relevant to the article, however, you brought it up in the article, so I think that makes it fair game.

    Assuming you had anything to do with this:

    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.

    I would like to know (1) why you would only accept a Framework donation contingent on them not violating your personal political beliefs (which have nothing to do with the purpose of such donations/ties), and (2) why you would say something so abrasive about other people in an article where you ask people not to be abrasive?

    If you disagree with my framing of question (1), that’s fine, but I would then appreciate if you explained how you see it from your point of view.

    For my own record, I have no love for Omarchy, it seems like a poor choice of ~~dotfiles~~ OS. But I have no issue with it or those who like it that would cause me to take issue with how *another organization* interacts with them.

    As for an on topic comment, I agree with your stance on AI and KDE, and I also don’t understand how this simple dialogue between maintainers turned into… this.

    Unfortunately this news reached me the same way as others, where I heard about the talk from the two pro-AI maintainers, and had almost none of the actual context until this post, which cleared up quite a bit.

    Like

    Reply

  12. Speaking as someone who happened to read the thread before the two people’s posts were taken off (and as a veteran of heated arguments online 🙂 ), I think it’s suboptimal to characterise that part simply as “began fighting with one another”.

    I appreciate that your original focus was on the need to talk about policy, and from that point of view, both “sides” of that interaction were noise. On the other hand, framing it only as “began fighting” is a form of neutrality as regards the content of the two people’s comments.

    (I realise the actual moderation decisions in that thread weren’t you btw.)

    To my mind, it is relevant that • Person A had been saying blatantly sexist and prejudiced things on Twitter. (although I don’t currently have a source from which to quote their actual words.)• Person B’s posts were suggesting that A wasn’t a desirable contributor to the conversation / the project, via flagging up A’s prejudiced words which indeed were contrary to the project’s values.

    Even though I don’t imagine you meant it this way, to sum up with no distinction between those “sides” carries the implication that {[horrible prejudiced comments]} is equivalent with {hey everyone, that person has been saying [horrible prejudiced comments], let’s take that into account}.

    For a lot of people, that misplaced neutrality is a warning sign: not of “bad intent”, but of the likelihood that nobody’s gonna have your back when the chips are down. (As per Desmond Tutu’s famous saying, “If you are neutral in situations of injustice, you have chosen the side of the oppressor.”)

    And to comment briefly on the actual moderation decision: I feel like if I’d been modding there, the most I would’ve aimed at B is something like “thanks for flagging, we’re gonna ban A, we’re gonna remove this subthread for clarity of original topic, in future we’d prefer to have suchlike reports via channel [wherever]”. Even if it’s inconvenient while it’s happening, doing some quite reasonable reporting via a dispreferred channel one time doesn’t strike me as a banworthy sin.

    my 2p, hope it’s helpful & doesn’t land as “one more person piling on”…

    Like

    Reply

Leave a comment