So you are saying all open source code that uses LLM code is infringing all the proprietary licenses of code that was also hovered up, is that a hill we want to die on?
The answer is obvious. If you want to protect your code, invent a new programming language that does weird things that make it difficult for an llm to learn.
Joking of course. Not sure copywriting code ever survives the AI era.
Essentially you're saying: AI companies are allowed to steal from all of us in order to make money, and in return they are allowed to read our private conversations and train future models in order to make even more money.
If humans are directing AI to generate code, it cannot be open source unless the human decides to make it. It is their vision that AI acted upon. So, i dont see the logic in this post or thinking. this sounds like an another altruistic progressive california thing
When AI becomes completely free to use -- that's when I'll support the idea that the result should be open source.
This is like saying programs should be open sourced because binary is machine generated.
Llms are the compiler, and there are skill levels with vibe coding.
It's hard to respond to something so reductive. This is the "everything's already been invented" angle.
Show us the math proof you have under your pillow. If there is not any, you do not have right to anything.
Well good news, I guess. AI generated code is public domain. It cannot be copyrighted.
RMS fans advocating for copyright enforcement... what?
Until a couple of years ago, learning was not considered theft.
Good idea. We just have to pay for the human-written part. Same with medical research and new medicines, new energy technologies, robotics, movies and music, and future AI. Eventually, as AI does more and more, everything will be free and open source.
You've figured out a way to eliminate money.
if i pay for AI to generate code, shouldn't i decide what to do with it?
by that logic, anyone who learned to code from open source projects should make their own code open source too.
I’ve been thinking a lot lately about how LLMs have essentially made all code open source in a strange and very capitalist way: They’ve appropriated our human work—without regard for any licenses—by force and without any compensation to the authors, and are now selling the code back to anyone who’s willing to pay for it.
But yeah, aside from that, I’m also releasing all my Vibe Coding games under the MIT license: https://github.com/retrogames23/
Following this logic, there should be no fees at all on coding models, since they scraped the web on this knowledge. Same applies for books, pictures, music btw. This debate has not even started yet.
This may be the path to the true public domain.
I could not disagree more with this view. Some of the works that reasoning models were trained on were already in the public domain. OP's statement is an attempt to get hold of code that they never build themselves. It takes human effort and human input - even if all it only determines is merely an initial random seed value - to generate the code in question. There is no fundamental baseline and there cannot be such a baseline that would provide for such a strict licensing of AI-generated code.
All code should be Open Source.
I see you and raise.. AI should be free
I think copyright law has simply failed to keep up with the reality of technological progress.
I don’t really see a problem with LLMs slurping up and regurgitating the world’s knowledge, if it was all freely and equitably available, and more importantly had been beforehand. We’d all herald it as an incredible innovation. And in some ways it is even now, because I do have access to it in numerous forms.
The problem is that some author’s livelihood is adversely impacted by this, as copyright is (in a sense) what ensured their remuneration for the original work. So LLMs undermine the existing social and financial order in deleterious ways. It’s the same thing we’ve seen play out in mathematics, a bunch if tech bros show up with a powerful tool, use it in a brash manner that likely causes real harm, then worse, start pulling up the ladder when big media conglomerates complain.
I suspect we’ll soon be in a situation where we no longer have easy access to unfiltered LLM outputs. If your OSS project was in the training corpus it will still retrieve it. It will not, however, retrieve content from NYT or other protected sources. The end result is that small-scale, solo creators are damaged, and large-scale creators are still afforded copyright protection.
There is no way to enforce opennness. The LLM gives me output, I iterate on it by hand and with other language models, who does it belong to now? We are ushering in a post scarcity world for a select few.
Nice pitchfork but I've got a hot take of my own:
No one goes to an LLM and says, 'Make code'. More often than not (and with exceptions) there's a very specific problem getting solved. The distinction between human 'effort' (whatever that means) and Ai effort (??) is more or less arbitrary... Infact, it's almost xenophobic and snooty. Pre-Ai humans had other humans build something like x but with y specific features. If it wasn't a problem then, it's not a problem now just because people don't have to directly deal with another human. In the same vein, I think people are forgetting that sometimes projects are birthed by slapping together code (or an instruction for code) from all across the Internet.
So really... What even is 'purely Ai generated' code? Nothing comes from nothingness and generally I'd prefer the purest expression of my ideas be protected regardless of who wrote it... Human4hire or Chat.
EDIT: This article is unnecessarily upsetting. I don't even understand the problem when at the end of the day. A shit ton of the value accrues to all of us... humans! Not even the companies (atleast not so far). If Artificial Intelligence really is a utility, like electricity, then wtf is exactly the problem? I hope nobody really gives any real thoughts to policing Ai usage... It'd be as unreasonable as policing a calculator or pen and paper.
Manners force me to stop here.
The Supreme Court wiping out trillions of dollars of equity value with a stroke of the pen by declaring all AI generated code open source would be hilarious.
And good for humanity.
It is disappointing to see many people who espoused that copyright is bullshit now saying copyright should make people do what they personally want. I suppose it was never about freedom for them, only getting what they want when they want it.
Whether programs written by LLMs should be protected by copyright law is also an entirely different question than should the authors be legally required to release the source code under a specific kind of license.
This article misunderstands copyright law -- citing an example that argues against it.
It cited Naruto v Slater, which unsuccessfully argued that the copyright for a monkey's selfie belong to the monkey, not to the owner of photography equipment.
The courts dismissed the case as animals cannot hold copyright.
---
While the courts did not rule it, the apparent copyright holder is the photographer Slater who received royalties.
This is directly analogous to AI code, in which a human did not directly author the characters but regardless had "significant creative contributions."
My patterns for loading data into Python is pretty much the same everytime with polars. I didn’t create that - I looked it up from a book somewhere. Do I need to open source it?
The AI machine just went out and read a bunch of examples then adapted them, which is what lots of programmers do.
If AI generated code is not copyrightable, releasing it under an open source license is meaningless.
At the end of the day you don't actually need to see the source code if you can just regenerate it yourself. If you use a later ai model, your code will probably be better anyway.
There is no such thing as AI generated code, just as there is no such thing as keyboard generated code. The only difference is that your keyboard manufacturer never had the gall to sign git commits -qwerty.
Is this for real? If you get your way gosh knows the amount of "proprietary" software that would be open-source. Also really by this logic, even 5 years ago if I copied a method and made it my own from stack overflow or got an answer from there my code should have been open source. Kinda silly no?
All LLM code should belong to the entity that hosts the model.
AGPLv3, not "Open Source." AGPLv3 is the most restrictive FOSS license, and it's one that other GPLs can be licensed into, but not out of.
Open Source licenses can be relicensed into anything, so they don't really count in the conversation - except that you'll need to include a copy of all the licenses touched that require that, and often include copyrights. You might be able to legally do that by just putting them on your LLM's website, though, who knows?
It would be very funny if distribution of LLM written software required that you include a milti-GB zip file of every possible Open Source license or copyright that may have been touched, and you were required to license it as AGPLv3. I honestly think that not doing so is a convenient reinterpretation of current law rather than an honest one; it's an absurd outcome, but not that absurd.
WOKE, no other word fits. You should move to another planet. Everything is stolen online, from marketing copy to books. And code is the one fight you want to have, the thing actually that is productive.
Every company on Earth that produces software is using AI to generate code. So this will never happen.
Open source people have had half a century to figure out how to make money with fairly little success (aside from a dozen or so very notable projects, and looking at all my AI slop, I don't see one of these hiding in there).
I don't particularly care about copyright or whatever ethical arguments yo may come up with -- I write code mostly for money, as do my competitors -- if you can figure out a way to stop them from using AI, I'm fine with not using it either. I'm fairly skeptical you can close Pandora's box though.
That aside, we may see more software going in a source-available direction -- either that or full cloud. Up until now being closed source was a competitive advantage, but now it's trivial to clone even fairly huge software suites, so that's less of a factor.
Having the source may also become a bigger advantage for regular users than before -- given they can now vibecode extensions/plugins, or customize the software more than what a config would personally allow. this probably wings the calculus away from a Photoshop-style license, and closer an Unreal Engine-style license.
Either that; or everything goes full cloud of course.
Both claims are easily rebutted .
> AI is trained on open source
It sure is, but the composition is a unique creation. Every creative artifact includes some derivation. A rhythm, instrument, media etc.
> AI isn’t made by humans.
Well neither is a canvas, paint, a chair. Everything we use is made by an automated tool, many products are nearly 100% automation.
Yes and, furthermore, all property is theft.
It already is. Just ask AI to give it to you.
Not if you see the AI-generated code as same as 90's java bytecode compiled by a tool (machine). In both cases, the input (AI prompts or 90's java code) is human-written. Just because a machine is involved in converting human input to some next low-level artifact, the low-level artifact does not become open source, when the human input is not.
Maybe by the same logic all code should be open source before AI if the author used stackoverflow, read a book, remembered their college education or read a blog?
The messy part in this idea is the “especially when human effort is minimal” bit.
No offense to anybody but I do believe there is no Intellectual Property (IP) anymore in the age of AI if one wants to defend or attack it was breached.
Code has lost almost all of it's value, who cares if it is open source. In a couple of years, you will say "make me a web browser" and 4 hours later performant, fully standards compliant web browser will be running.
The argument between open and closed source has little value anymore and will soon not even make sense to discuss. It will be like writing "2 + 2" and arguing over whether my "closed source" copy of "2 + 2" vs your "open source" copy of "2 + 2". When something is trivial to reproduce it becomes absurd to argue over who owns it.
Hot take indeed, and something I disagree with. If I am paying for AI, my code is my code, no?
And everything that can be handled as a common should.
That is everything bar one own mind and body, I guess, but your lineage might vary, and for sure some people out there would rather be deemed owner of everything under the sun, starting with human corps and souls.
Yes.
Until and unless every creator and contributor of every piece of training data consumed by these models is fairly compensated for the use of their works, then absolutely everything created by these models should be 100% open source - including the models themselves.
All code should be open source.
I agree
Honestly, I'm cool with that if all AI models are also made open-weight.
What’s the point, if you can generate infinte code?