Open Source Projects Banning AI, From QEMU to NetBSD

Added: 2026-06-01

[00:00:00]While the Linux kernel is quickly becoming vibecoded, that's a real thing I did I didn't have on my 2026 bingo card. Uh if you look at the number of commits coming into the Linux kernel that are written by AI LLMs, it's it's exploding at an amazing rapid rate where the majority of commits will be assisted by AI sometime this summer. It's wild. While all of that is happening, at the same time, a large number of other prominent open-source projects are going in the opposite direction. They are completely banning all uses of LLMs completely. Let me walk you through just a couple of them here.

[00:00:48]Uh, the Haiku operating system, the open- source re-implementation of the B operating system has banned all AI contributions to their project, including code, documentation, and translation. In fact, the leadership of Haiku OS also appears to be discussing expanding their AI ban to include uh prohibiting vibe coding of any software in their repository.

[00:01:18]And they're even they're even talking about uh forbidding and banning AI generated posts even just just in their forum. This is a this is a thing that's happening. I mean you can't take a much harder stance against AI than that. Or can you? Here's a few more projects. Let me just run down some of the most some of the most intense ones from my point of view. QMU I mean QMU the ubiquitous emulator. um has now has a policy that says quote uh that their policy is to decline all in caps their their usage of capitalization here. Decline any contributions which are believed to include or derive from AI generated content here. Let me show you a screenshot of it right there. Uh it's pretty pretty hardcore anti- AI. Uh there's NetBSD, which I've talked about in the past, but AI generated code. In NetBSD is quote presumed to be tainted code and must not be committed. This is part of the official NetBSD policy. Uh do not commit tainted code to the repository. Uh code generated by a large language model or similar technology such as uh GitHub, Copilot, uh etc., etc., etc. um is presumed to be tainted code and must not be committed uh without without prior written approval.

[00:02:49]In other words, don't don't don't submit any any AI generated code. Uh the Zigg programming language has a total ban on using AI in any way. And I mean in any way. I I want to read for you some of the ways that AI usage is banned with Zigg. This is part of their official code of conduct. Um, no LLM generated content, whether it be code or pros, no paraphrasing LLM generated content. So, not only can you not submit something the AI generates, you can't even paraphrase what it said. Uh, no LLMs for editing, including fixing spelling or grammatical errors. No LLMs for translation. English is encouraged, but not required. You're welcome to post in your native language and or rely on others to have their own translation tools of choice to interpret your words.

[00:03:48]No LLM for brainstorming and then sharing the results of that brainstorming, even if you create the pros. No LLMs for finding bugs. No, no talking about using about Sorry, let me let me let me start to say this one over. No talking about use of chatbot LLM services.

[00:04:14]Zigg has banned talking about using AI systems.

[00:04:21]Yeah. I mean, you thought the haiku ban was rough. Uh Zigg went hardcore. Um OBS Studio OBS Studio the you know the live streaming compositor tool um quote code must be human written is an they have an AI machine learning policy AI machine learning systems are prone to generating plausible sounding but wrong code. They said um all text and content submitted to our project which which includes code description issues and comments must be human written. the use of uh GitHub copilot and other assistive AI technologies is heavily discouraged because they require humans to write it.

[00:05:08]I mean this is it's interesting seeing the stark contrast where you've got uh the Linux kernel and the Linux foundation going allin on AI. Red Hat going allin on AI. Their business model is now AI. They're uh providing bonuses and incentives, financial incentives for their employees to use AI more and create more a AI features. And then you've got Haiku, NetBSD, Zigg, Opia Studio, QMU, and there's quite a few others as well that are that are going anti- AI to the extreme.

[00:05:45]That to me is is is is an almost startling thing when you see a the open-source world which is historically already quite fractured is fracturing in previously almost unimaginable ways. the the amount of of extreme differences here is is going to be very very stark especially when you start to consider that so many of these projects are interdependent on one another and they're they're taking very very firm stances against AI code generation. Um, now Lionus over in the Linux kernel world, Lionus Torvolt recently posted, actually just uh about 4 days ago, 5 days ago, that he was quote, "Not entirely happy about the flood of AI code changes coming in late in the development cycle for the current release of the Linux kernel." Quote, "This is the heads up that I'll be pushing back on pointless pull requests with the fixes that just aren't that important." And yes, several of these series were uh were triggered uh by AI code review. He posted this over to the Linux kernel mailing list uh where he clearly is is unhappy with the significant amount of AI generated code late in the development cycle. He's talking about uh unnecessary churn this late in the game. um and talking about how the current release candidate for the for the current version of the Linux kernel uh is uh pretty bigger than than RC's have traditionally been and you know he's not happy about that. So, we'll see realistically going forward if Lionus's unhappiness with some of the AI problems that have occurred, some of the code churn and some of the just huge number of code commits coming through has any impact on how how quickly the the number of those AI generated code commits continues to grow because the Linux Foundation is clearly behind that.

[00:07:55]Uh they're they're all for it. Uh Jim Zlin, the executive director of the Linux Foundation, did a video with Anthropic, the AI company, just uh just a couple weeks back talking about how amazing it was and that was the future of Linux and open-source and all that sort of thing. And we also see a lot of other Linux projects, you know, Linux companies like Red Hat, uh, projects like Fedora, Omari, and so many others just fully embracing the the AI generated code, uh, to the point where so much code is being generated by AI that they could not possibly be fully reviewed by humans. It's not possible. the the amount of code that's being generated for the Linux kernel, for Omari, and for so many others, they could not be fully completely reviewed.

[00:08:52]There's just too much of it. And so there's there's a lot of that happening at the same time as all these other projects saying we won't allow any any AI code. One has to wonder if there's maybe there's some sort of a good middle ground.

[00:09:07]May maybe there's a happy middle ground where people use AI systems maybe a little bit to like help them out on especially difficult bugs, but otherwise write the code themselves. You know, maybe maybe that makes sense. Maybe we need to get some sane heads in here and not go to the extremes. I don't know. Or maybe maybe one of these extremes is right. Uh I I you know me, I'm somewhat skeptical of the of the AI side of things. I see a lot of concerns and problems and potential pitfalls with mass adoption of of AI code generation.

[00:09:43]Uh just a whole bunch of them. In fact, maybe maybe I should do a a whole show where I talk about some of those. But uh right now it's it's fascinating watching the differences between all the various projects right now. Um, oh, oh, I want to say thank you to all the subscribers because lifetime subscriber wall number seven, the Solaris wall is now full. We now have seven full retro themed walls filled with the names of amazing lifetime subscribers to the Lunduke journals. Just huge numbers of you. We just started these walls just this last December. So, I mean what, six months ago. and uh they've just been filling up at an almost everinccreasing rate. Wall number eight will be debuting in the next couple of days and I'll tell you right now, it is Windows 1.0.

[00:10:43]Uh because why not? It is the version of Windows with arguably the worst possible color scheme. I mean, Windows 1.0 almost makes your eyes bleed. the color scheme is so bad on it. So, why not make a lifetime subscriber wall of Windows 1.0?

[00:11:02]So, if you want to get on the Windows 1.0 wall and you're already a lifetime subscriber, go to lunduke.com. There's uh contact information on how to get a hold of me, get a hold of me and say, "Hey, add my name. Here's how I want my name to be displayed. Add my name to the Windows 1.0 wall, and I'll get you I'll get you on there." Um there's already a couple of names that will be sliding into that wall from uh from the from the last couple of days. But if you want there's plenty of spa plenty of space on the Windows 1.0 wall and if you want to get a new lifetime subscription the massive discount on that is good through May 31st. Uh so again, go to lunduk.com.

[00:11:44]Scroll down just a little bit and there's a little link there that says uh you know, hey, massive discount blah blah blah blah blah. And you can go and grab a massively discounted lifetime subscription, which literally is lifetime. Gives you all the perks, gets you on the lifetime subscriber wall, helps support the work of the Lond Journal. Truly fantastic. And that's good through May 31st. Afterwards on June 1st, the price just goes back to normal. Very, very simple. Go check it out. Thank you to all the subscribers who make it possible to do the work that I do. Could not do this without you. And with that, ladies and gentlemen, boys and girls, nerds and nerds across the inner tubes, I do declare end broadcast.