Recently, there has been a surge in slopcoded new/hobby “operating systems”. Such slopcoded projects – which, due to the nature of “AI” tools, effectively consist of stolen code – will not be featured on OSNews and submitting them is fruitless.
Other websites may choose to employ lower standards, as is their prerogative, but OSNews will not. I obviously cannot guarantee nothing will ever slip through the cracks, but I will take utmost care to ensure OSNews remains free of these so-called “sloperating systems”. Plagiarism, license-washing, and code theft have no place in the world of enthusiast and hobby operating systems.
Good on you!
Seconded
Oh god, what will they do now?! Have mercy on their souls.
Pax vobiscum.
If this site had been around the 70s or 80s, there would be an op ed about refusing to deal with any OS projects that use compiled code… LOL
It’s funny you think that “AI” is on the same level as compilers.
yes that.
the idea that what LLMs do to write code is any different than what humans do to write code (ie, we look at other peoples code) is just pure bullshit.
no one works in a vacuum anymore Thom.
Especially when preached by one who admits never have written any software at large.
LLMs don’t write code, they output text. Unlike a person, it can’t reason.
Parodper,
Have you actually tried though? Obviously it’s not the same as consciousness or free will….but today’s LLMs do have logical reasoning. Some of the newer models actually let you debug/view the reasoning stream. I’m not going to claim it’s super human or anything like that, but it seems likely LLMs do reason as well as average humans. I could paste some here but honestly it would be best if you saw it for yourself.
Treated lick a black box, the reasoning is there. On the other hand there is a double standard for humans versus LLMs.
Of course, I agree. And I want to congratulate you on coining the term “sloperating systems” as well.
Is it really *that* much different than being inspired by, and having read, other OSS operating system’s code?
I agree in that I would much rather read about 100% “hand-crafted” projects, but this is because I have that romanticized view of the bedroom hacker pouring all their heart into their project.
The more I think about that, the more I come to the conclusion that it is not that different if AI tools are used. Maybe the “hand-crafting” just happens at a higher level then.
I guess I have not fully made up my mind on that subject 🙂
I really don’t get the current trend of virtue signaling through reactive anti AI positioning.
This is especially obvious in the usual lack of understanding around code reuse. In real SW development, functionality is rarely created in a vacuum. It is usually derived from, inspired by, adapted from, or built on top of previous work. That is how SW evolves. Reuse, compatibility, shared conventions, and incremental improvement are normal parts of engineering, not evidence of creative failure or ethical deficiency.
> I really don’t get the current trend of virtue signaling through reactive anti AI positioning.
It seems to have turned into a highly emotional affair where you have to chose a side (or will be burned on a stake a an heretic).
War time story: I am a heavy Arch/AUR user because I am forced to install a lot of proprietary software, which often depends on old software blobs only available as AUR package. So of course the AURpocalypse worried me. I do understand that as an AUR user I have to read every single script before installing it — but I would not even have known every attack vector out their in the wild — so I needed help.
Out of desperation, I built https://github.com/manticore-projects/aurscan simply to get another, more educated opinion before installing anything. Since it worked for me, I thought sharing would be nice.
And then all hell broke lose: Thumbs down, rude comments and insults everywhere — yet not one single qualified criticism on the code or the technical approach itself.
At the same time, after 20 years of developing software, it has become my most successful publication with 100 stars within 3 days (which is great, considering what niche I am serving to).
@Thom: The Ottoman Empire was a center of knowledge, science and wisdom (paralleled to the Chinese empire) — then banned the printing press for Arabic-script texts for roughly 250 years after Gutenberg (under penalty rules dating to the late 1400s). So while Europe mass-produced knowledge, the Islamic world remained limited to hand-copied manuscripts.
But I am sure, you are the purest and most enthusiastic of all of us, kudos!
Cytor,
,
A lot of people are angry about AI, and there are plenty of legitimate concerns, However there are a lot of weak arguments going around too. I think these may be having the opposite effect on people who become “unconvinced”. One such argument is equating AI training to stealing, most FOSS licenses do not objectively support this assessment. If anything most FOSS licenses have aimed to enshrine the right of downstream users to use the code as they please without restriction. This has long been touted as a feature of FOSS and not a bug! And yet anti-AI voices would like to have anti-AI restrictions get retroactively added to existing FOSS software.
For better or worse though most FOSS licenses are permissive and allow code to be used for AI/LLM applications. The actual point of contention, insofar as traditional FOSS licensing goes, is that derivative works be properly licensed.. IMHO we need more focus on enforcing FOSS license terms.
We should remember that the rights granted by FOSS licenses are there on purpose and protect our stake in FOSS and give us the freedom to decide how we use it.
You are spot on (as always), but even outside of the realm of free licenses: people went to libraries, read books and gained knowledge which they applied commercially — at least at my times this was how things worked (different age, different universe). The whole western civilization is based on knowledge and wisdom copied (“stolen”!) from Islamic/Ottoman scholars. Suck it up and deal with it …
> Is it really *that* much different than being inspired by, and having read, other OSS operating system’s code?
exactly. it’s not different at all – it’s just faster.
Lets be honest: often even better, because there is no laziness anymore to cut corners and “fixing this properly later” (when later never comes).
I have the romanticized idea of ambitiously made AI pouring memory and cycles into creating something which people thought it could never create.
So what if AI learns by being inspired by other developers? Don’t we all?
Thank you.
The difference between AI slopcode and Open Source is just that. Slopcode is theft. Open Source properly attributes.
I think you may be confusing plagiarism with theft.
Glad to hear it.
Though It’s not really any change in principle for the site.
Are you virtue signalling? It’s unbecoming. You seem to be reacting to “stolen code” and tossing out hot button terms like: slopcoded, “AI”, lowering standards, sloperating systems – in order to make an emotional appeal to your users.
What if… AI-assisted hobby OSes were created using Open Source and everything was properly credited. What if the licenses were public domain and other highly permissive licenses that required no attribution?
I am not sure exactly what you are reacting to. Is it llm-assisted code or the theft itself. Those are two different things. What if a code-generating AI was trained only on public domain, highly permissive licenses that required no credit?
I think it might be time to pull my osnews sponsorship – thom, you are preaching from ignorance. it’s mid 2026, go learn what you’re actually railing against and actually understand how people wrote code before and after LLMs.
I get it – c++ sucks because object oriented is bloat, and we should all write in assembly…..
it’s not stealing code any more than anyone else does.
done. patreon canceled.
Thom, I endorse your commitment to avoid giving attention to “sloperating systems”, and I have decided to support you via patreon. Keep up the good work!
Does this mean you will no longer be covering news from Microsoft given their preference for slop code in Windows, and I assume the rest of their portfolio?
The Linux kernel also has a policy for AI submissions, does this count as slop code and can we still expect coverage?
I thought this too; however, if you re-read what Thom said, he’s clearly talking about hobby projects.