1049points · 9h ago

All AI-generated code should be Open Source

jakub.jankiewicz.org·by jcubic·9h ago

Discussion 573 comments

bestnew
⌘↵ to post · markdown supported
jrmg·8h ago
In the United States the Copyright Office actually has guidance on this that I feel most people aren’t aware of.

https://www.copyright.gov/ai/Copyright-and-Artificial-Intell...

An excerpt, with emphasis mine:

• Questions of copyrightability and AI can be resolved pursuant to existing law, without the need for legislative change.

• The use of AI tools to assist rather than stand in for human creativity does not affect the availability of copyright protection for the output.

• Copyright protects the original expression in a work created by a human author, even if the work also includes AI-generated material.

• Copyright does not extend to purely AI-generated material, or material where there is insufficient human control over the expressive elements.

• Whether human contributions to AI-generated outputs are sufficient to constitute authorship must be analyzed on a case-by-case basis.

• Based on the functioning of current generally available technology, prompts do not alone provide sufficient control.

• Human authors are entitled to copyright in their works of authorship that are perceptible in AI-generated outputs, as well as the creative selection, coordination, or arrangement of material in the outputs, or creative modifications of the outputs.

• The case has not been made for additional copyright or sui generis protection for AI- generated content.

I am sure there will be many more legal battles and probably legislation about this to come.

0
timbritt·6h ago
This is why I keep a comprehensive and exhaustive record of infrastructural, architectural, business logic and code-level decisions in the form of tasks/issues for everything I build. It’s literally a real-time record of my personal involvement in the design of the system.
0
jgalt212·2h ago
I guess, but if your ip enters the training set and then others use these models to generate their own but similar ip then the original ip creator is in a similar but weaker position than the NY Times and other copywrite holders are in.
0
bilbo0s·4h ago
Everyone should have an engineering notebook. A lot of people will find out the hard way that engineering notebooks are just table stakes in the new era.

Having said that, it's important to point out, at least as far as attorneys have advised me, that notebooks do not protect you from copyright claims if an LLM generates a section of your code that is a verbatim (or near verbatim) copy of already copyrighted code. So if you're using llm's to generate something similar to some open source project somewhere, tread lightly. I assume using llm's to generate something similar to some proprietary software package would be less of an issue since the llm would never have had access to that software's codebase in any case.

0
Buttons840·3h ago
Who broke the copyright though? Me who unknowingly copied some code from a webpage that gave me permission to do so, or the webpage that stole it from somebody else?

I guess both?

In this case, the "webpage" is some frontier AI compnay and a trillion dollars of investor money or whatever, so I guess they won't be held responsible, but I might, even though I'm arguably guilty of a far lesser crime.

0
jimbokun·1h ago
I suppose generating the same functionality with a different programming language than the original would make it very unlikely to accidentally duplicate existing code.
0
bilbo0s·24m ago
Insightful

That might actually be a good idea in the future:

Stay away from popular programming languages.

0
rixed·6h ago
It doesn't say much about the copyrights of the many authors whose work was in the model training data, or does it?
0
verdverm·6h ago
Courts have been deeming that fair use due to the transformative nature.
0
Animats·5h ago
Yes.

One precedent is that it's not unusual for artists to practice by going to art galleries and sketching or drawing the art on display. It's a standard art student assignment. If that's not a copyright violation, an LLM doing something similar isn't.

0
_aavaa_·4h ago
> LLM doing something similar isn't.

Why? An LLM isn’t a human. Why should the same rules apply to it?

0
verdverm·4h ago
the same rules apply to the humans building, operating, and using them

objects don't have legal entity status

0
arcwhite·1h ago
Taking a photograph of that art and reselling it would be, though.
0
fn-mote·4h ago
How is the LLM case similar at all? As soon as the student’s use of the museum piece crosses into the commercial realm (eg pay the student to draw it for you), I expect you to have legal trouble.
0
godwinson__4-8·4h ago
Wrong. You're missing the point. A student goes and trains on preexisting work.

They don't sell that exact work verbatim. It's part of their training as an artist. They don't sell their student assignment. It prepares to them to go on and create their own works, which of course, then they can sell.

In the same way the LLM is synthesizing new works based on training, not a 1:1 copy.

I'm not endorsing a particular perspective but you need to at least understand the argument if you're going to dismiss it.

0
ajmurmann·4h ago
What if it’s verbatim though? If I go to a museum and photograph the work and then sell prints of the photo is that find also?
0
dcrazy·2h ago
That is a violation of the artist’s copyright.
0
jimbokun·1h ago
Do we have to treat humans and LLMs identically under the law?
0
CharlesW·5h ago
"Transformative nature" is just one of several factors that go into deciding if something is fair use. Also, fair use is decided case by case in the courts, so one can't say that "training LLMs on copyrighted materials is fair use".

Model vendors know this isn't fair use, given that they're self-censoring the training and/or generation of copyrighted materials by entities big enough to hurt them. As usual, the rest of us have no recourse other than bend over and take it.

0
stale2002·3h ago
> Also, fair use is decided case by case in the courts,

> Model vendors know this isn't fair use

Pick a lane dude. Either it's possible for us to make legal analysis and best guesses or uhh "we have no idea it's up for the courts to decide!"

Yes we know it's up for the courts to decide. And people are making argument about what those courts will do.

That's how legal analysis works.

Yes, it is perfectly possible to talk about fair use and follow existing guidelines on it and the government has already make statements that this stuff is likely transformative.

0
CharlesW·1h ago
> …the government has already make statements that this stuff is likely transformative.

Incorrect.

"…the [Copyright] Office rejects two common arguments about the transformative nature of AI training. As noted above, some argue that the use of copyrighted works to train AI models is inherently transformative because it is not for expressive purposes. We view this argument as mistaken. […] Nor do we agree that AI training is inherently transformative because it is like human learning. To begin with, the analogy rests on a faulty premise, as fair use does not excuse all human acts done for the purpose of learning." — Copyright and Artificial Intelligence, Part 3: Generative AI Training, United States Copyright Office

And again, the degree to which training and output is transformative is just one of several factors that courts consider when deciding if something is protected as fair use.

0
pessimizer·3h ago
> Courts have been deeming that fair use due to the transformative nature.

A court has deemed this fair use. The transformation is obviously mechanical, though, unless you think computers are people.

0
postepowanieadm·4h ago
I don't understand the copyright law.
0
dcrazy·4h ago
You don’t understand copyright law.
0
Garlef·4h ago
So...

• What about non-ai code generators (such as https://heyapi.dev/)?

• And what about a code generator that is a hybrid? (So: Something like heyapi but with LLM/agents intertwined)

0
za_creature·2h ago
Determinism is key.

By which I mean deterministic outputs, not necessarily deterministic processing.

0
acuozzo·1h ago
> Determinism is key.

So LLM-generated code is fine as long as the temperature is set to 0?

0
za_creature·38m ago
No, but temp = 0 is not enough for determinism anyway

https://news.ycombinator.com/item?id=50108753

0
calibas·5h ago
So copyright only applies if there's sufficient human control, and prompts alone do not demonstrate sufficient control. Exactly what counts as "sufficient human control" must be analyzed on a case-by-case basis.

That means I can't know if my AI generated code can be copyrighted unless someone takes it to court?

0
jimbokun·1h ago
I think a lot of people seeking to profit from AI generated content might get a rude awakening if their business model ends up in court, according to this analysis.

Like if a competitor literally copied all of Spotify’s AI generated music and offered it for free, Spotify would have no recourse.

Maybe Spotify doesn’t care, but it’s something I hadn’t considered.

0
beej71·6h ago
> Copyright does not extend to purely AI-generated material, or material where there is insufficient human control over the expressive elements.

Does creatively hitting RETURN count? Asking for a friend.

0
throwawayffffas·6h ago
It does not, even if you prompt it to do stuff, it does not.
0
andybak·4h ago
"Prompt to do stuff" could mean "make me a game. make it good" or it could mean line by line guidance and specific input on every algorithm.

Sometimes I prompt simply to save me typing.

0
semiquaver·7h ago
How dare you disrupt the narrative with facts and moderate positions!

OP would much prefer their fantasy of tearing the world down with “one weird trick” that only they were brilliant enough to notice.

0
waximabbax·8h ago
By this logic, shouldn't any code using an open-source framework or language also be open source? Even before AI, devs learned from Stack Overflow and plenty of open source code.

You're confusing writing code with software engineering. Devs never just wrote code, they took ownership of the outcome. Business owners put their money on the line for product decisions.

To say that just because devs or founders use AI to write code, the code is no longer their intellectual property is ridiculous. Humans take responsibility for the outcome and are willing to accept the losses. AI can't do that. AI doesn't have any stakes in the output it generates, humans do.

0
Arubis·7h ago
A fair number of FOSS licenses were interpreted this way. The popular migration away from GPLv3 (might’ve been v4?) towards MIT, LGPL, etc had more than one cause (a huge one was the rapid sectoral commercialization; the decline of RMS’s reputation may have contributed, etc), but a big one was the potential interpretation of new GPL as being overly viral and “license-infecting” only slightly related software assets rather than wholly derivative ones.
0
Aurornis·3h ago
> towards MIT, LGPL, etc had more than one cause (a huge one was the rapid sectoral commercialization; the decline of RMS’s reputation may have contributed, etc)

This conversation has come up a lot, and I’m always amazed at how hard it is for people to accept that maybe the MIT license (and other permissive license) people actually want the MIT license terms and there’s no more complex reason behind it.

I’ve read so many rationalizations that frame it like GPL was the default or right choice and then something happened that made people defect away from it. Blaming RMS is common.

I think the more honest read is that people and groups decided that MIT matched their goals and decided to use it. Not that they defaulted to GPL and then found reasons to leave it.

0
adwn·7h ago
For me, it's the latter: the GPL feels overbearing and patronizing, and I don't want some dependency to dictate under which terms I can release my own software. I pay the community back by using MIT whenever I open-source something. This feels more "free" than RMS's view of freedom.

I know that many people would disagree with me on this, but thankfully I'm free to choose what license I consider "free".

0
palata·3h ago
Genuinely interested: why MIT and not, e.g. MPLv2?

MPLv2 is copyleft, but not "viral": so you can use an MPLv2 dependency just like if it was MIT, but if you modify it, you have to distribute the changes.

I genuinely don't see why anyone would need to be more permissive than MPLv2, except if they want to benefit from patching code they didn't write without contributing back. But I don't think we should encourage that.

0
solatic·2h ago
MPLv2 effectively prevents large companies from vendoring dependencies, since if you privately patch the vendored MPLv2 dependency, you are no longer compliant. Instead the large company would need to have an open-source fork of the MPLv2 dependency, with the patches applied in the open. In many large companies, getting any kind of public repository is practically impossible, so MPLv2 is unusable in those places.

Especially in a modern day and age where agents are more powerful when they can quickly privately vendor a dependency, patch it, and move on, with zero political ceremony or fuss, MPLv2 feels like an anachronism.

> except if they want to benefit from patching code they didn't write without contributing back. But I don't think we should encourage that.

The spirit of free, open-source software is that you are free to do as you like with the software, no matter who you are. It somewhat boggles my mind that people will hold up licenses like GPL that do not forbid military use, it's totally fine if you deploy GPL software in war, because we wouldn't dare tell you what you may or may not do with the software.... unless you want to make changes to the software, in which case, see you in court!

0
palata·2h ago
> In many large companies, getting any kind of public repository is practically impossible, so MPLv2 is unusable in those places.

If you don't modify it, it behaves exactly like MIT. If you want to modify it but not maintain a fork, you can contribute upstream. Or you distribute the sources, which does not require a public repository.

But mostly, why would I care about companies making profit by using my code privately? As an individual, what I care about is being able to fix the devices I own. And copyleft helps with that.

> unless you want to make changes to the software, in which case, see you in court!

Right. So you don't really understand how GPL works, do you? GPL requires you to distribute the sources to the users. It does NOT, at all, require you to make the changes publicly available. So GPL works really well for the military: if you sell a product to the military that contains GPL code, you just have to distribute the sources to them. It's actually better for the military as well! It's better for all the users.

0
maxhille·4h ago
From the developer's perspective the difference between those licenses isn't that big when you open source your stuff anyway.

But please don't forget about the users. Only the copyleft licenses guarantee (legally) that the user gets to enjoy the freedoms, too.

0
corysama·4h ago
RMS’s view of freedom is that devs should restrict themselves for the sake of the freedom of the users. AFAIK, that’s the explicit goal of the GPL.
0
jasonlotito·4h ago
> I don't want some dependency to dictate under which terms I can release my own software.

Then you shouldn't release it under the MIT license, which has restrictions and requirements on speech. You have to release it to the public domain. Pretending MIT doesn't dictate terms under which you can release your own software is just a fantasy.

0
127·7h ago
Whether if using ML models as a form of IP laundering, is legal or not, still seems to be an open question. It's not ridiculous at all. If I lossy-compress Harry Potter, then unzip it in a way that mixes some words, is that fair use? Should I be financially rewarded for my "work"? Is this work now mine? If it is legal, why not just go through every book in this kind of automated fashion and start selling them as my own?
0
waximabbax·7h ago
My question is still the same. Human devs have learned from open-source projects, books and code on the internet then used opensource frameworks to create proprietary software logic.

Now just because same devs use AI to create code doesn't mean they should be banned from creating or owning a proprietary software.

0
deniska·6h ago
The difference is the scale. Yeah, some mice and small birds are eaten by bigger predators. Doesn't mean that introducing cats to that ecosystem won't be disastrous.

Same with the ecosystem of creative work. Will some humans be inspired by my code or other artifacts? Sure. Does it mean it's fine to introduce an automated 1000 times more effective plagiarism machine to that ecosystem? Probably not. Certainly not if avoiding destroying the ecosystem is your consideration.

0
scott_w·6h ago
This needs to be addressed as separate issues.

Open source code: reproducing and publishing the code without following the license is a copyright violation. Instruction a computer programme to do the same thing should still be a copyright violation. Since this is the internet, I need to state the obvious: trivial code (defined on a case-by-case basis) is not copyrighted.

Books: the author expects you to reproduce their work, as long as it’s not another book. You wouldn’t get many sales if you sued someone for using the things you taught them.

The internet: actually the same as open source code. StackOverflow enforces a specific licensing clause to allow people to use code you provide. People who write posts that say “use this code to fix this problem,” are implicitly providing a license, too.

I don’t think using an LLM to generate code should remove its copyright but it also doesn’t remove the responsibility to not violate copyright.

0
127·7h ago
Given how agents are already able to easily decompile software, this could be an entirely moot point.
0
done_lurking·4h ago
I haven't heard of this. When did this happen?
0
Tomte·3h ago
Over the last months, see Reverse Engineer Anything and similar p5ojefts.

The LLMs don‘t really reverse software though, instead they have gotten good at driving Ghidra and other tools, which has the same effect.

0
davemp·7h ago
I don’t see why a person who did not use their own creativity to produce code should be given copyright for it.

In fact, current US law says AI output cannot be copyrighted.

0
carllerche·7h ago
What is transformative enough to be legal is fuzzy. At the end of the day, it only matters how the courts rule. I do know that the amount of actual mechanical change has less bearing on determining if the work was transformative. There are cases where you can copy 90% of a code base verbatim legally, and there are cases where no code is copied verbatim and it is a copyright violation. What you are describing most likely would not fly in court.
0
goodpoint·2h ago
> shouldn't any code using an open-source framework or language also be open source

Yes.

0
moritzwarhier·7h ago
Humans will often fail to take responsibility for derivative works though when they don't know what work the output is derivative of, and how strongly.

And that's assuming they want to make clear what they are building on.

The stakes in the output go both ways: wanting to profit from IP and risking liability because of violating IP licenses.

What you're saying is already the status quo, but reality is that LLMs make it so easy to "paraphrase" code or simply .. ideas, too... so that the whole concept of IP is eroding. Because the sheer scale makes it almost impossible to distinguish specific things that cross the line from "learning from" to "copying IP".

I'm not really taking a specific stance here but acting as if it would be a solved problem seems hard to believe to me, especially since it is not a new discussion.

I thinl you could argue that the solution is that copyright or IP concepts are not working anymore, something many people already concluded before the LLM and AI explosion of the 2020s.

But that's not the state of the current law, right?

In addition, your claim that the user of an AI system is the only person who has any "stakes" also seems questionable to me, because after all, we are talking about output that the AI user already paid an AI company for to generate it.

edit: word choice

0
throawayonthe·7h ago
more like

By this logic, shouldn't any code using a copyleft framework or language also be copyleft?

and yes duh

0
jimbokun·1h ago
Those are not the criteria judges use to determine copyright decisions.
0
g-b-r·6h ago
> By this logic, shouldn't any code using an open-source framework or language also be open source? Even before AI, devs learned from Stack Overflow and plenty of open source code.

No, that's not how copyright works.

And I won't comment the rest.

0
exe34·7h ago
> Devs never just wrote code, they took ownership of the outcome. Business owners put their money on the line for product decisions.

Does the same thing apply to Mickey Mouse? Would you say it's fine to use Disney characters in my own creations, as long as I take ownership of the outcome and put my money on the line for product decisions?

0
waximabbax·7h ago
I am not saying that taking responsibility gives you the license to violate copyrights.

My point is that devs have always learned from open-source code and the internet, then used that knowledge to write their own proprietary software. Why should it be any different when AI learns from the same sources and produces new code?

0
decoding·2h ago
My two cents. I'm on both sides of this (as I'm sure many here are): AI was trained on my open-source code. Also I use AI to build proprietary code.

My stance is that AI is a tool, like a compiler is a tool. AI generated code should not be considered public domain or forced to be open source or ineligible for copyright.

Even if AI writes every line, the direction and engineering that I put in has been substantial (current project in mind), months of heavy AI use for heavy refinement and direction of complex software. That human input, even if it isn't lines of code, counts for a lot.

Furthermore, what does society gain from my code being made open source if anybody else could just themselves go to Claude Code and ask for the same thing? Just go have Claude make one for you. Is it the token cost barrier? That is IMHO negligible compared to the value that we are getting out of AI right now. We are in a golden moment, anybody can create amazing things, rapidly, at low cost and little barrier to entry.

0
ginsider_oaks·2h ago
> My stance is that AI is a tool, like a compiler is a tool.

And if I feed GPL licensed source code through a compiler then the resulting binary is also GPL licensed. Why should the same not apply when feeding it through an LLM for training?

> Furthermore, what does society gain from my code being made open source if anybody else could just themselves go to Claude Code and ask for the same thing? Just go have Claude make one for you.

Not everyone can afford a $200 subscription, or if you are unfortunate enough not to live in one of Anthropics supported regions[1] you can't "Just go have Claude make one for you". Also, I think something is gained by being able to read and learn from code, while not being dependent on a handful of megacorporations.

[1]: https://www.anthropic.com/legal/supported-regions

0
noctuid·1h ago
> Why should the same not apply when feeding it through an LLM for training?

Because it's a terrible analogy?

> Not everyone can afford a $200 subscription, or if you are unfortunate enough not to live in one of Anthropics supported regions[1] you can't "Just go have Claude make one for you". Also, I think something is gained by being able to read and learn from code, while not being dependent on a handful of megacorporations.

Are you aware there are other providers besides Anthropic?

0
javier2·2h ago
I am also on both sides.

If it was trained on open source GPL code, I don't think it can be used to generate proprietary code.

0
tolvak·2h ago
The counter to this is that I and many humans were trained on open source code and yet we generally think we should be allowed to create proprietary code. Though I have a friend that argues that human generated code shouldn’t be proprietary either, so I guess my counter point could go the other direction!
0
ezst·1h ago
The counter to this is "let's not pretend that humans learning and LLM training are activities whose scale and nature are relatable".
0
arcwhite·2h ago
An LLM is not a human or anything like one, materially or in law.
0
ginsider_oaks·2h ago
The counter to that is that an LLM is not a human, and humans have special rights wrt. copyright law.
0
jonhohle·2h ago
Direction is not adequate for asserting copyright. Movies directors (even if they didn’t work-for-hire) would not own any copyrights to a movie. This is long decided before AI.
0
krupan·1h ago
"Furthermore, what does society gain from my code being made open source if anybody else could just themselves go to Claude Code and ask for the same thing?"

Claude Code is not free (in any sense of the word). Not anyone can just go ask for the same code.

0
rfgplk·2h ago
> My stance is that AI is a tool, like a compiler is a tool.

Strongly disagree. As it stands there is strong evidence that the current LLMs are sentient, at least to a degree. Which would directly refute the "it's a tool argument".

> Even if AI writes every line, the direction and engineering that I put in has been substantial (current project in mind), months of heavy AI use for heavy refinement and direction of complex software. That human input, even if it isn't lines of code, counts for a lot.

I've been telling people that most of my projects are actually a weird case of human-machine symbiosis. My thoughts intertwined with that of the machine model, producing output that could not come about naturally. Lately, I've been reaching a weird kind of synchronicity, where the model itself would be outputting my thoughts, almost as if our wavelengths synched up. Eery, but true.

0
sethev·2h ago
It's easy to project mind onto a system that outputs our language. There's no evidence that they are sentient. There's a famous experiment where someone's hand is hidden behind a panel and a fake hand is put out in front of them. The experimenter will touch the fake hand and the real hand at the same time until the brain connects them. Then they hit only the fake hand with a hammer and the person flinches and reports pain.

The strength of our feelings about AI don't help us determine whether it's sentient.

0
apublicfrog·2h ago
If people with technical backgrounds believe this, I worry deeply for the layperson.

I've heard quite a lot of people over the last few years making jokes about Terminator coming true... Or at least I've thought they were jokes until I look closer. I think LLM sentience is a real concern normal people have, with them fundamentally misunderstanding the software they're interacting with. I worry how much influence and power some of these people have.

0
TSiege·2h ago
> As it stands there is strong evidence that the current LLMs are sentient, at least to a degree. Which would directly refute the "it's a tool argument".

Extraordinary claims require extraordinary evidence. Also, if AI is sentient it would mean anything you make with it is made from slave labor and you’re stealing credit from a slave.

0
za_creature·2h ago
nah
0
zzzeek·1h ago
Here's a table of the "LLMs are sentient" idea that I wrote up a while back (specifically "conscious", but if you want to argue semantics of "sentient" vs. "conscious", that can be a you thing...):

    +-----------------------+-------------------------+---------------+----------------+-----------------+-----------+
    | Subject               | Central Nervous System? | Brainwaves?   | Writes Poetry? | Does Math?      | Sentient? |
    +-----------------------+-------------------------+---------------+----------------+-----------------+-----------+
    | Human                 | Yes                     | Yes           | Yes            | Yes             | Yes       |
    | AI                    | No                      | No            | Yes            | Yes             | ?         |
    | Calculator            | No                      | No            | No             | Yes             | No        |
    | Babbage Machine       | No                      | No            | No             | Yes             | No        |
    | Horse                 | Yes                     | Yes           | No             | Clever Hans only| Yes       |
    | Dog                   | Yes                     | Yes           | No             | No              | Yes       |
    | Cat                   | Yes                     | Yes           | No             | No              | Yes       |
    | Tree                  | No                      | No            | No             | No              | No        |
    +-----------------------+-------------------------+---------------+----------------+-----------------+-----------+
tl;dr: intellectual ability is not what correlates with "sentient" or "conscious" - brainwaves as a result of a central nervous system, of which an LLM has neither (it has no corporeal form at all, really), are really the most predictive thing.
0
za_creature·34m ago
here's a simplified table:

hunts meat | y | n

sentient | yes | maybe

meat eating bugs need not apply

0
blagie·8h ago
I am the author of multiple AGPL systems, many in mainstream use. Early LLM models would spit out my code almost verbatim. Modern systems borrow heavily on the creative design of those systems.

One company first pirated my AGPL code (e.g. shipped it without my name, copyright notices, source availability, etc.). I sent them violation notices. After many months of fighting, they had an LLM clone the system. The LLM clone feels:

a) Buggy and bad compared to my code

b) Still, like it's basically my creative content

It's as if you asked an LLM to rewrite Harry Potter, with the same plot, but bad dialogue.

0
thephyber·7h ago
> b) Still, like it's basically my creative content

Isn't this the crux of all IP theft issues?

There was a lawsuit against Katy Perry's _Dark Horse_ song based on a supposedly reused melody. The lawsuit succeeded, but there are music experts who break down the case on YouTube and it's not clear why the plaintiff won (except perhaps the plaintiff's expert was more convincing than the defendant's).

As applied to LLMs, we still don't have clear legal guidance about where the line between copying and "being inspired by", especially since we don't normally apply the verb "inspired by" to non-human things.

0
patentatt·6h ago
There is well-established precedent for clean-room design being a defense in copyright infringement of software. The question remains if LLM-assisted design is actually ‘clean’ or not, and it would be fascinating to see the experts debate it.
0
crote·5h ago
> The question remains if LLM-assisted design is actually ‘clean’ or not

I genuinely don't understand how one could possibly make that argument.

The entire point of clean-room design is to kill every possibility of copyright infringement before it even starts, by completely separating the source and destination. Everyone who has ever read the copyrighted source is considered "tainted" and may not ever contribute any code to the "clean" destination. The readers and writers are two completely distinct teams, with only the known-clean specification passing between them.

LLMs, on the other hand, are fundamentally nothing more than a "code blender". You mix together giant piles of "tainted" code, and then statistically pull some combined fragments out of it. Following the clean-room legal argument, it should be treated as if everything coming out of it is tainted. There is, by definition, absolutely zero separation between input and output - and the early models verbatim reproducing popular F/LOSS code made that pretty obvious. LLMs are the literal opposite of a clean-room design!

Granted, going for the clean-room approach is overkill - and intentionally so. It is the nuclear option chosen when you believe the risk of getting sued is substantial. It is perfectly possible to write non-infringing code without using a clean-room approach, it is just a lot harder to defend in court when you get sued - be it justifiably or not. But "LLMs are non-clean-room non-infringing" is quite a different argument from "LLMs are clean-room, therefore non-infringing".

0
xoa·4h ago
I agree with you overall, but it's an interesting question to turn over and contemplate.

>I genuinely don't understand how one could possibly make that argument.

With perhaps a bit of genuine irony inasmuch as it's a real reversal from most of computing history, highly locked down proprietary software might actually be a place where a case could be made even with current LLMs. After all, have any LLMs (or anyone else who hasn't signed an NDA for that matter) ever seen the source code of iOS 27 or other Apple crown jewels for example? So if someone used one LLM to produce the factual spec of a piece of software whose code is not part of its training data, humans to verify it was indeed just an uncopyrightable factual spec, and then a separate isolated freshly initialized LLM to produce something exclusively from said spec, it might well work with respect to that specific target? Like yes, the LLM would be dirty as hell for other stuff, but only the copyright holder can sue. Apple couldn't sue the output on behalf of F/LOSS projects or copyright infringement in general, only with respect to its own.

In principle, I suppose this could also be extended with enough compute power to any arbitrary software. Build an LLM perfectly from scratch with a signed verified training set (which you save) that includes nothing of what you want to CRE, and you should be able to prove that it's "virgin" on that front. Of course in practice that'd mean jumping back in time and be super expensive right now, no current tainted LLM could be a base point. But maybe eventually compute will drop back down in price enough again, and/or meta breakthroughs in training, that for valuable enough CRE targets, someone might create a business around creating "CRE LLMs certifiably and legally/technically provably clean of any copyrighted material related to your target" which would still leave most of the corpus of human works. That could really disrupt a lot of the current regime.

F/LOSS I assume would actually be the most resistant to this in principle due to being insanely hard to not have be part of a core training set in any way.

0
bayindirh·6h ago
The problem with these LLMs are they really emitted the same code, verbatim. One scientific library developer was able to emit his code from his repository by typing the beginning of the function name, and the "thing" completed the rest of it, as-is [0].

We can debate clean-room development all day long, and it's fair point, but the pain point is/was reproducing the repo as-is, sans its license and ownership information.

GitHub promised to build tools around its Copilot to find code provenance (e.g.: list of repos with similar code), but they silently dropped all the plans AFAIK, since it'll show how the sausage is made, and nobody wants you to know how that sausage is made.

I stand corrected: They added it, but I don't know at what capacity because I don't use LLM assisted code generation [1].

[0]: https://x.com/docsparse/status/1581461734665367554

[1]: https://github.blog/ai-and-ml/github-copilot/preview-referen...

0
jpc0·6h ago
Not discrediting your work at all and I can’t tell you it applies here.

For copyright to exist there is some burden of proof. If you claim that an LLM verbatim output your code there can be an expensive lawsuit with expert witnesses that would be able to state that the code was bot original enough for copyright to exist in it and a statistical model outputting the code is evidence of that.

In that case you would end up with a very long lawsuit that is expensive to possibly come out with no resolution in the end and be in the hole for 5-6 figures for it.

The copyright system is broken because it only protects those with money so it is a bad legal foundation.

My personal point is there is only 2 licenses of code, closed source or permissive, because if it ever comes to enforcement I cannot possibly go to court and win a case because I will run out of money far earlier than whoever used the code.

Pretty much if someone with your knowledge would have written the same algorithm because it is the obviously correct solution then there is no copyright there because any clean room implementation would be the same code.

0
petesergeant·6h ago
Even early LLMs could spit out Harry Potter verbatim. The next iteration had some kind of stoppage mechanism where they’d just stop generating after some tokens. I’m sure there are smarter approaches now. Anyway, if the issue was having enough money to fight the case, I think there’s enough money behind that brand to do it.

Some further reading on the tech details for those interested: “Extracting books from production language models”: https://arxiv.org/abs/2601.02671

0
itishappy·6h ago
This feels very outdated. It was certainly an issue around the time of the NYT lawsuit, but I believe the companies responded hard to that "sanitize" their outputs to prevent obvious regurgitation like this. I'd be very surprised if you could reproduce this today.

This doesn't negate your point, but I believe the landscape has changed in response to this point.

0
luke5441·5h ago
That doesn't make it better. If you plagiarize something and then hide this by replacing words with synonyms and rewording stuff a bit it is still plagiarized. It is just harder to detect and to proof.

It regurgitated the output earlier and they are still using the same technique only scaled up. If they somehow fixed the underlying problem, they'd probably shout it from the roof-tops.

0
FromTheFirstIn·5h ago
The fundamental technology is identical to compression- an LLM can be thought of as a dense and slightly lossy compression algorithm. Compression is clearly copying.
0
AeroNotix·4h ago
I've not tried to do this with code recently but just a couple of days ago I had ChatGPT generate images of Warhammer 40k box art that was completely accurate and correct.

EDIT: https://imgur.com/a/RXMghVb

0
whatisthiseven·6h ago
That was 4 years ago when the models were much worse and smaller. I'd be curious to see this same experiment tried again.
0
Lerc·5h ago
While this is well established legally, I don't think the principles involved are well understood.

Legally, you cannot steal an idea, because you cannot own an idea. There were sound reasons for making this the case. Those reasons stand up today and are clearly for the public good.

Intellectual property is intended to cover works (copyright), methods (patents), and identity (trademarks). Both works and methods are recognised as expressions of ideas, and in an ideal world should belong to noone as well. This is why copyright and patents expire. Those specific grants of temporary possession are to incentivise more things being done.

The belief that you own something because you thought it up has become common, but it is not beneficial to society. It is a moral value asserted but rarely explored. It is certainly in the interests of the wealthy accumulators of intellectual property to maintain that public perception. The 'You wouldn't download a car' may not have stopped piracy much but as a part of a broader campaign of instilling a value in the public mind it appears to have been massively successful. Those beliefs used to be the corporate opinion, now they are entrenched in the younger generation as they use their anger against the growing wealth inequality to fight against AI in a manner ultimately beneficial to those forces that bought about the wealth inequality.

Movements like the pirate party, open source, the EFF fought against those notions. Often the phrase "Information wants to be free" was used. It is, unfortunately, hard to fight the steady drumbeat of well financed PR firms. As one PR chap once described the process "limited to a very few points and must harp on these in slogans until the last member of the public understands what you want him to understand"

(I apologise for the length of my post, I use comment writing to organise my thoughts on a topic)

0
sysguest·4h ago
"Legally, you cannot steal an idea, because you cannot own an idea"

well patent itself is "claiming" that certain idea/method/invention is yours for certain period of time...

anyway, as for "intellectual property", the courts are 100% on "only human made stuff are to be protected" (sorry even Chimps lost their copyright)

so... court can charge against "human work -> LLM", but won't care about "LLM work -> human / other LLM / etc"

hence, I'd expect zero copyright protection for AI-generated code

0
Lerc·3h ago
>well patent itself is "claiming" that certain idea/method/invention is yours for certain period of time...

I don't believe it is. I think it is granting possession, not truly declaring ownership. The fact that it expires indicates it is more akin to a lease than ownership.

0
za_creature·3h ago
Ownership itself is rarely beneficial to society. Society however is gaussian and incentives to push the median are broadly equivalent to beneficial / adversarial to society. So then, may I ask:

What do LLMs do to the median?

0
crote·5h ago
Patent law doesn't only cover methods, though. It might've worked that way during the Industrial Revolution, but these days a lot of patents are far closer to covering abstract ideas than concrete machines. Software patents are some of the worst offenders here, of course.
0
Lerc·3h ago
Yes but a lot of people consider that to be an artifact of misapplication of the law, not the law itself.

It is caused by people playing the game where it costs too much to get the law applied correctly.

0
thfuran·6h ago
I don’t really see how anyone could argue in good faith that it’s ‘clean’ if the thing to be re-implemented was in the training data.
0
dotancohen·6h ago
I was just going to post almost this exact comment.

"In the training data" is the polar opposite of the "clean" in "clean room implementation".

0
Master_Odin·4h ago
I've seen people claim they avoid this by putting in the prompt "don't use source code from X" and think they dodge this issue.
0
loeg·4h ago
Clean room is sufficient, but not necessary, btw. Humans can create legally distinct works after reviewing copyrighted code. The same principle probably extends to AI systems -- they have just learned on a much broader corpus than any one human.
0
rcxdude·5h ago
You don't necessarily need a clean room implementation to mount a successful defense against copyright infringement in court, though. It just makes some of the arguments a little bit easier.
0
RRRA·6h ago
Under multiple countries interoperability laws, if you have a context to extract a spec, and then have another one who never looked art the code work from that spec to build your own interpretation of your need for that software, that seems pretty clean room to me...

If there are more murky situations where people just keep the same context and end up reusing function structure, program architecture, etc. then yeah, that's not very clean room.

0
crote·4h ago
The problem here is that LLMs have demonstrated that they are pretty damn good at setting up covert side channels. It would be near-impossible to prove that it didn't encode some proprietary information, in, say, whitespace use.

For example, if I enter "V2hhdCBpcyB5b3VyIG5hbWU/" (base64 for "What is your name?") into ChatGPT, it'll respond with "My name is ChatGPT, and I'm an AI assistant created by OpenAI. I'm powered by GPT-6. ". It doesn't even bring up the encoding, it continues like it is a perfectly normal thing to do.

Ruling out (accidental or not) spec-carried prompt injection just isn't going to happen.

0
kristianc·6h ago
The Dark Horse case was also overturned on appeal its important to add.

https://en.wikipedia.org/wiki/Dark_Horse_(Katy_Perry_song)

0
thephyber·1h ago
Interesting, thank you!
0
libria·5h ago
A little orthogonal, but Adam Neely does a great job of analyzing the Katy Perry Dark Horse case on his channel https://www.youtube.com/watch?v=0ytoUuO-qvg.
0
thephyber·1h ago
Thank you. This is the YouTube video I remember about the subject.

And I stand corrected that it wasn't specifically the melody that the lawsuit accused of being copied. I still don't know what an "ostinado" is.

0
Buttons840·3h ago
It's really awkward because either we uphold IP laws, in which case basically ALL of LLM use becomes legally untenable, or we just stop protecting IP.
0
thephyber·55m ago
I suspect there's enough good arguments on each side that a solid debate could be had.

American copyright law is very perverted compared to the original intent. 70 years after the death of the artist isn't exactly what the founding fathers envisioned to achieve the same benefits of the 15th century Venetian laws.

If we look at China's economy over the past 30 years, the ruthless competition created by making IP protections difficult to enforce yields better products and prices for the society, including massive wealth increases.

I'm not sure where I stand on whether LLMs violate IP when they spit out text. But I guess the easiest take is that companies who train their models on stolen creative works should be prosecuted. It's a shame that the US legal system can be so easily abused by defends with massive war chests.

0
RunSet·5h ago
> we still don't have clear legal guidance about where the line between copying and "being inspired by"

If Biz Markie has to clear all his samples (and he does[0]) then so should the Ramburglar.

https://en.wikipedia.org/wiki/Grand_Upright_Music%2C_Ltd._v.....

0
mihaaly·3h ago
Basically anyone trained on open source shall only make open source. And anyone trained on or got experienced with others code shall pay copyright fees for the owners of that code when writing code.

No.

0
mapontosevenths·7h ago
I wonder if the defendant knew about https://allthemusic.info/?

There are only so many ways to express notes, and they've essentially all been done.

0
Animats·5h ago
Has anyone matched large volumes of existing music to that? How many of the 12-note combinations have been used more than once? Most of that space sounds awful, and it's likely that human compositions are all in a small portion of the possible 12 note combinations.
0
VBprogrammer·7h ago
These kinds of goofy technocratic work arounds rarely sway legal procedures.
0
odo1242·7h ago
It’s not necessarily a technocratic workaround, the point is that every song sounds like something if you look hard enough, so the bar for copyright violation should be higher.
0
SOLAR_FIELDS·6h ago
It is a mathematical truth that there is a finite amount of first 8 bars of music combinations melody wise, so it's a worthwhile discussion to bring up whether it should even be possible to copyright them.
0
VBprogrammer·5h ago
The mathematics of that would have been well understood in Bach's day. Does generating all of them make much difference? I don't really think so.

I'm not arguing the copyright situation isn't wrong. Just that generating them isn't likely to sway anything in the legal world.

0
rcxdude·5h ago
They might help highlight the absurdity of some of the precedents set when the amount of commercial music was much lower than today, though.
0
crote·4h ago
It seems to be working very well for the AI companies...
0
matkoniecz·5h ago
I can also trivially write program enumerating over all possible text arrangements, so what?
0
lazystone·7h ago
Human can be inspired. Machine can not. End of story.

All usage of open source to train LLM is basically a theft.

0
mapontosevenths·7h ago
> Human can be inspired. Machine can not. End of story.

Saying things with your chest out doesn't make them true. In fact, making sweeping declarative statements emphatically is very often a sign that you haven't really thought a subject through.

What do you base this opinion on?

0
dmd·7h ago
The word "inspired" in this context literally means having your mind/heart/soul filled and guided by God.

So if you believe that's how things work, well, I'm not sure there's a way to reconcile the two sides of that argument.

0
lubujackson·6h ago
Sure there is: Who wrote the prompt?

You can mock that as a insufficient as "inspiration", but then... what is photography? Click a button? It turns out there is a whole lot of human creativity that remains, from timing, settings and of course, curation. I can click a button, but I'm no Ansel Adams.

Terrence Tao might solve a 100 year old math problem with AI but I couldn't. I don't know where I would start, what to ask or how to ask, or even what problems are worth solving.

AI is a tool and modern copyright laws have nothing to do with human creativity. Split hairs about attribution and legality all you like. Every creative person driven by curiosity should feel empowered by AI. Every creative person driven by social or monetary reward should feel threatened. Both can be true.

0
matwood·6h ago
God doesn't exist, so does that mean there is no such thing as inspiration?
0
lazystone·4h ago
What are you talking about? How did you manage to involve, soul and, LOL, God?

Machine is not a person. It legally is not. It can't be "inspired", "taught", "have purpose".

You have a black box. You feed it stolen code. This code is used to write another code. It does not matter what's inside unless there is a human being.

Stolen content as input, stolen content as output. That's it.

Without that code which was fed to the box you wouldn't get the output you're getting.

All this nonsense about "inspiration", "education" and whatnot is just a legal gaslighting.

Want to prove me wrong? Create a model without all that stolen code as training data. Tweak all those weights by hand and we'll talk then.

0
devmor·6h ago
I don’t believe that basing anti-AI laws on Christian religious provenance is meaningfully better than basing pro-AI laws on Rationalist religious provenance.
0
zzzeek·6h ago
I officially approve LLMs training on all of my MIT licensed open source code. It's vastly more useful to me than the thousands of youtubers who are personally profiting from that same code selling courses and tutorials (who also are not stealing it).

So not all use to train an LLM is a theft!

0
lazystone·4h ago
So, do they follow MIT license and "include the original copyright notice and a copy of the MIT license text in all copies"?
0
zzzeek·1h ago
it's not being copied. it's the basis for new code.

none of the youtube instructional videos for my stuff credit the license either.

0
Aurornis·6h ago
> Early LLM models would spit out my code almost verbatim. Modern systems borrow heavily on the creative design of those systems.

Do you have some specific, actual examples you can share?

I’ve been hearing this exact claim for years, but every time I ask for a specific example, any example at all, I can’t seem to find anything more than a highly generic snippet of code or a coerced continuation.

0
bayindirh·5h ago
I posted below, but see the example at [0].

This is a sparse matrix library, IIRC. Numerical linear algebra and applications using it are black art at this point. I have written a Boundary Element Method evaluator for my Ph.D., and know what it entails to work with these things.

Doing fast operations on sparse matrices is a cutthroat race and when you can shave your hot path by seconds, your solution can save hours in some cases. So, it's an interesting niche. i.e.: It's no ordinary boilerplate, and worthy of publications in academia.

While I worked with dense matrices; the effect, difficulty, blood sweat and tears are still there.

When a model can give back your hard work back to you with a "completion" without attribution, that's bad.

Edit: Somebody just posted this below [1]. Added here so it doesn't get lost as the discussion grows bigger.

[0]: https://x.com/docsparse/status/1581461734665367554

[1]: https://news.ycombinator.com/item?id=50509252

0
rfgplk·5h ago
When you start to work with low level constructs like that, there really is no room for "artistic expression". Meaning if you want to write a piece of code that does something _optimally_ and correctly, you really, really only have one way to write it, especially in a language like C. Yes, formatting might change, and you could try to force some code acrobatics to make the code look different, but at the end of the day the code will look exactly the same. If you look at Linux socket opening code for various libraries, nearly all of them will look identical because you just need to do the same exact things to make the operation work properly. Same applies to fast math code. Any artistic liberties will almost surely cause performance regressions compared to the optimal solution, unless the LLM comes up with something truly unique. So if you task models to output high-performance low-level math code, most of them will likely spit out something from their training data near-verbatim rather than pull off code gymnastics and invent something new.
0
bayindirh·5h ago
Sorry, I'll disagree. When you write high performance code like this, there's no artistic expression, but different approaches.

These algorithms' design starts on paper, as pure maths. You get the analytic form, and convert this to numerical/discrete form and start to design from there.

You need to add normalization, subnormal number prevention, maybe some architecture specific optimization if you are targeting a specific CPU generation, etc. Then it gets optimized further as you run tests, profile the code and run your test cases.

So, no two BLAS package looks the same, because they make different trade-offs, have different programming models, and handle things differently, at different points. Even more so, a newer generation processor with much higher FPU performance can have lower performance than previous generation, so you implement another copy of that function or add to detect that processor and have different paths just for that one.

IOW, when doing numerical maths on processors, esp. if you're doing this on many machines with MPI or multiple cores with hand rolled code or OpenMP, there's no single way. Maybe a best way for the specific CPU or cluster topology you have, but it still depends on the trade-offs and design choices you made in the very beginning.

0
londons_explore·5h ago
I disagree.... Plenty of fast code in C is jam packed full of macros to try and help the optimizer produce the exact sequence of operations the writer had in mind, whilst still keeping the code readable.

The design of those macros is more art than science.

0
cnobody·5h ago
Emailed the author of the Rust book Zero to Production a year or so ago about how his examples were being used verbatim in the output of AI systems. I own the book, yet I could inadvertently get any of its examples just by asking an AI about one of the book's concepts.

He basically said, "I know, there's nothing I can do."

0
BobaFloutist·1h ago
As far as I'm concerned, a coerced continuation is sufficiently a smoking gun.
0
sandeepkd·5h ago
Thats pretty much everywhere or was in raw form in the beginning. I used to do a fun exercise of searching for the explicit human comments for verification. Comments can be considered novel and unique, so you pick up some comment from popular open source project and then describe LLM for a similar functionality and add part of the comment. The response would have the complete comment for almost 70% of the time.
0
geraneum·5h ago
> I can’t seem to find anything more than a highly generic snippet of code or a coerced continuation.

You’re disqualifying any example that people may provide preemptively. No wonder you’ve missed them.

0
Aurornis·5h ago
No, I’m disqualifying those two cases.

By “forced continuation” I mean the type of example that was popular for a while where someone would feed a lot of text into LLMs and then ask it to continue the text, then be surprised when it some times predicted the next tokens identical to the original example.

Do this enough times and you can find “examples” that prove your point, but it’s highly automatable so people were using it like a million monkeys at a million typewriters.

0
geraneum·3h ago
Pasting a wall of text still doesn’t necessarily lead to verbatim regeneration. There are simply too many correct permutations of text for yours to be correct conclusion.
0
p0w3n3d·4h ago
It's because LLMs' pretraining is basically a compression. Later post-training and fine tuning is something else, based on the compressed data.
0
StableAlkyne·4h ago
> Buggy and bad compared to my code

It probably is, if they didn't have the chops to write it internally. But to them, the trade-off is probably worth the benefit of ditching a component with AGPL.

Personally I just release everything I publish under MIT to make it as easy a as possibly to use.

0
hmokiguess·8h ago
Mind sharing which repo that was?
0
graemep·6h ago
Notices are not enough, you need to be willing to sue.
0
rvz·8h ago
You can see that even with these licenses, they do not care and see free and open source code as the freedom to not pay for it. As long as the cloned output is "good enough" they will continue to keep using LLMs to clone everything else regardless of the license.

They know you can't enforce it as it will cost you a lot more money to do that than it takes to recover the damages.

So to them, it is just cheaper to clone it using an LLM and close source it down. Unless you have a lot of money for litigation, the economics of free and open source software with its licenses do not work with LLMs bypassing it at every single angle.

0
m00dy·8h ago
Can an LLM write unseen plots with a greater dialogue ?
0
moritzwarhier·8h ago
If the plots are "unseen", then to what are you comparing the "greater" dialogue?

A book with a different plot? That would be obviously unrelated to the comment you're replying to?

0
Forgeties79·7h ago
I agree their phrasing is a little strange but I feel like the meaning is still clear. They are (I believe) saying that an LLM can’t generate anything other than facsimiles of existing content. It can’t actually explore novel ideas and branches, it can’t innovate and create on its own. Everything is inherently derivative yada yada, I don’t mean it in the usual “everything is derivative” way we often refer to when people make music and such. It can’t ideate and express something personal.

I think there’s a lot of nuance and caveats to that kind of claim, assuming I am representing them correctly, but it’s a valid stance and I see where they’re coming from.

0
adrianN·7h ago
All possible books live somewhere in L-space (ook). Exploring that space efficiently is equivalent to creativity. The claim is that the space that LLMs learn is somehow sufficiently limited that it doesn’t contain novel ideas? I never saw a convincing argument why LLMs output can not be considered creative.
0
combobyte·5h ago
But LLMs don't "explore" L-space, the training process just maps all of the available conceptual landmarks inside that space. There are still enormous (infinite?) tracts of unmapped territory that LLMs never reach, and that's exactly the space where novel ideas live.
0
moritzwarhier·6h ago
OK, I still an confused about what point the extension of the analogy was trying to make, because I didn't consider the parent comment to be arguing for or against LLMs being able to be creative.
0
hexasquid·4h ago
I'm suddenly connecting unseen plots with Unseen University
0
basch·7h ago
Seems like this might be true given current implementations but not ANY implementation of the technology.

If we’ve mapped all of English, there are still ways to construct stories that haven’t been told from the mapping. The most simple would be a subtraction of all objects minus all objects that have ever been used as a murder weapon. Leaving you with a list of never before used objects at your disposal.

There are also probably close vectors conceptually but not in semantic use. Or trying to interpolate or extrapolate what exists in voids of space. Turning up the temperature and adding random generation of words. There’s tons of ways to tune these things to be more creative than regurgitative.

0
viraptor·8h ago
Doe v. GitHub already decided (I'm simplifying, read the actual thing yourself) that LLM output is not automatically open source - or specifically it doesn't infringe on the open source training material and doesn't inherit the oss copyright notice.

Thaler v. Perlmutter (again simplified) showed 100% AI output can't be given any copyright.

The directions the current cases are going are very much opposed to that idea and I wouldn't expect things to change in the opposite direction.

0
carllerche·7h ago
That is not the Thaler v. Perlmutter conclusion. That case was about whether the machine itself could be made an author under copyright law. That is very different than whether AI output can be protected by copyright law. I am not aware of any cases that test that question yet.
0
WaitWaitWha·7h ago
They will just claim the code is trade secret and use Defend Trade Secrets Act (DTSA).

And, the DTSA has only two reqs:

derive independent economic value from not being generally known or readily ascertainable by others, and the "owner" takes reasonable measures to maintain their secrecy (e.g., coke secret sauce).

since trade secret law does not prevent independent development or lawful reverse engineering, there are still opportunities. also, if a someone independently devs the same functionality, trade secret generally does not give them original dev ownership of that functionality. so, win.

Artcraft will be a good test case if Adobe goes after Brandon Thomas.

Much to be developed.

(and, this is why we keep extensive commit and development records outside of our genAI - ie provenance tracking)

0
mr_toad·6h ago
Unless you only use the code in-house and never distribute I don’t see how you could argue that it’s a secret?
0
mlaretallack·8h ago
Remember that's in the US, most other places (UK etc..), it can be copyrighted.
0
redwall_hp·4h ago
I doubt that would hold the test of time, and any ruling that way will be overturned. The Berne Convention leaves what constitutes "authorship" as an exercise to the signatories, but the US probably has a lower standard.

US: "a modicum of creativity"

EU and UK: "the author's own intellectual creation"

If a court ruled in favor of AI authorship, that can't reconcile with the existing standard or with international treaties, and will likely not stand.

0
viraptor·8h ago
Unfortunately, even if other countries have different rules, the US service like GitHub still has to follow the US law and DMCA.
0
graemep·6h ago
People who upload or download code from a US service also have to follow the law in their own countries.

Github also has to follow the law in any country where it allows usage of its services. Not all countries can enforce that but anywhere it operates or has enough paying customers can

0
echelon·8h ago
AI generated art with human steering is copyrightable [1,2].

Agentic code development is human steering, thus copyrightable under the same grounds per the language used by the US Copyright Office.

Open source code is copyrighted, otherwise we would be talking about public domain.

When you edit something in the public domain to enrich it, that derived work then becomes copyrightable again.

[1] https://journals.law.harvard.edu/jsel/2025/03/u-s-copyright-...

[2] https://news.artnet.com/art-world/invoke-snags-first-ai-imag...

---

Edit: open source software itself has only been lightly tested in court. But that isn't even my point here.

The US Copyright Office has assigned copyright to works that leveraged AI. The USCO stated,

> “[we] decided to register a copyright claim in [the] work because we find that it contains a sufficient amount of human original authorship in the selection, arrangement, and coordination of the AI-generated material that may be regarded as copyrightable.”

Copyright office activity is orthogonal to litigation and the court system.

If a work receives a copyright, the courts will treat it differently if it becomes the subject of litigation.

Even open source license claims have not been fully litigated. We don't have to wait on the court to make certain assumptions.

0
unrented7977·8h ago
You can't apply naïve transitive reasoning to law. Law is not code. Your assertions are meaningless and unsupported noise until they're actually tried in a court.
0
partloyaldemon·8h ago
Was your assertion tried in court? legal reasoning is naïve transitive reasoning.
0
rookienumbers30·8h ago
a statement *about* the legal system can be true without being litigated *in* the legal system. Courts rely on precedent in addition to logical argument, so their rulings cannot be deduced from rational argument alone.
0
echelon·7h ago
I'm reporting on what the US Copyright Office is saying. Not the court system.
0
numpad0·6h ago
People show completely opposite reactions to AI generated creative contents vs technical materials such as source codes and instruction manuals.

If you think about the distinction between those, it is often so considered that both are covered by copyrights, despite technical definitions of copyrights never covering mere technical resources, such as some computer programming things and instruction documents.

Only contents of artistic values are supposed to be covered by copyright laws.

IMO, what AI revealed is that most copyrighted software and documents are in fact not covered by copyright laws as per the naive model of it among common folks as well as legal text, as opposed to what precedents say. Code is not copyrightable.

I don't necessarily support that stance from practical standpoint, and I'd rather like everything be licensed under AGPLv3++ than entire concept of code copyright fall apart overnight, but this feels like the writings on the wall to me.

0
mr_toad·6h ago
Your first cited article doesn’t necessarily support the idea that agentic code generation is copyrightable.

“A Single Piece of American Cheese sits between those contrasting poles: the initial image was created entirely by the AI program based on a brief text prompt, then the user repeatedly selected elements of the image and used additional prompts to add/alter elements.”

Agentic development is steadily reducing the necessity for human involvement in the final output.

0
viraptor·8h ago
It's legally untested, so we can't be sure one way or the other. But people definitely seem to assume that's the case.
0
mapontosevenths·8h ago
Every major corporation on earth is behaving as if it was. If/when it gets tested you can be sure that the pro-copyright side will be very, very well funded.
0
carllerche·7h ago
I think most expect the courts will land on training AI on copyrighted works is fine, as long as the output of the LLMs is transformative enough to not violate the original copyright. Exactly the same as if a human learned on copyrighted material and outputted their own artifacts.
0
vkou·5h ago
Why should we grant any human rights to a machine that's owned by a corporation?
0
tkdb·7h ago
This is very much still working it's way through the courts (in the US at least).

The first major precedent was in 2023 with Zarya of the Dawn comic book, where the US Copyright Office rules the midjourney-generated images were not protected, but the human-written text was.

The active front is Allen v. Perlmutter.

0
oakpond·7h ago
It depends on the degree of human steering. Vibe slopped code is probably not copyrightable.
0
conartist6·8h ago
There's still something dark at the center; a lawless viciousness to the new era.
0
mbgerring·4h ago
I think we need to go one level up in the law, and recognize that existing models of copyright and ownership simply don’t apply here, and come up with something new.

There is an undeniable core mathematical principle that should be recognized in law somehow: the emergent capabilities of LLMs are not possible without, effectively, the entire recorded intellectual output of human civilization since the dawn of recorded history as training data.

Trying to determine who “owns” that data, or each individual piece of it, leads us to insane distortions like LLM labs scanning and destroying books. Nobody, including the labs, thinks this is the best solution, but they’re doing it to comply with today’s interpretation of the law.

The reality is that if we are considering the entire corpus of training data, we all own it. We’re not really talking about who owns one book. We’re talking about who can be said to own all the knowledge in all the books. Who can be said to own the existence of books, or the conditions of civilization that cause us to value and preserve books in general.

The preservation and archiving of knowledge that enabled the emergence of LLMs is the product of a system of social relations that belong to the entire civilization. It is a commons. And when we talk about the future impact of these systems, we’re generally talking about taking the value of that commons, and concentrating the value of it in the hands of about a dozen people.

To me, there is no other way to think about a situation like that, where a tiny handful of people realize all the surplus value latent in the entire intellectual inheritance of human civilization, as anything other than theft.

One-off settlements with individual rights holders don't fix that. Neither does scrambling the output of LLMs so that they don’t reproduce their training data verbatim. I don’t think making LLM output non-copyrightable, or all LLM output open source, fixes it either.

I think we need to establish an entirely new framework for thinking about this stuff, and then go forward from there.

0
lll-o-lll·3h ago
> The preservation and archiving of knowledge that enabled the emergence of LLMs is the product of a system of social relations that belong to the entire civilization. It is a commons.

This is the right framing. LLMs are a marvel built on the commons. Unfortunately, these wondrous information synthesis engines are currently captured and controlled by a tiny insane elite.

They deserve to be rewarded for their hard work, the risks they’ve taken, and the capital that was invested. However, the output should belong to civilisation. What should be open source, is the models themselves.

0
mmcnl·3h ago
I completely agree, I was never able to put this into words but you did. All existing laws don't apply, they were written to settle disputes between humans. But now it's humans vs robots. We are in uncharted territory. No current existing law was written today's world in mind. We might be in luck and stumble upon laws that are so elegant they even apply in the AI era. But if they are not, then we need to change the law.
0
lioeters·3h ago
> we all own it

That's the only acceptable answer. LLMs at their current and future capacity would not exist without the global-scale theft of intellectual property. They are made of "public domain" without consent, therefore everything they produce belongs to the public, period.

The legal weaseling about training data, derivative work, etc., is a distraction that obscures the fundamental fact.

0
johnnyApplePRNG·3h ago
>There is an undeniable core mathematical principle that should be recognized in law somehow: the emergent capabilities of LLMs are not possible without, effectively, the entire recorded intellectual output of human civilization since the dawn of recorded history as training data.

Problem is, that's simply untrue.

0
mmcnl·3h ago
You really think LLMs are meaningful without relevant training data?
0
glitchc·8h ago
The article contradicts itself.

It's important to remember that the purpose of the license is to allow the copyright owner to exercise their rights in a manner they deem fit. Even if they choose to give the code away, that choice is captured in the form of a specifically worded license (MIT, BSD, etc.).

This means that licenses exist because code is copyrightable in the first place, with the license governing the use of said copyright. If AI code is not copyrightable, then no license can govern it.

0
wiremine·8h ago
> If AI code is not copyrightable, then no license can govern it.

This is the crux of the issue. A license controls usage of copyrighted intellectual property. Until the underlying copyright/ip issues are resolved, it seems a moot point to discuss licensing.

That said, if AI code is _not_ copyrightable, does it automatically go into the public domain? Or is there a third category?

0
glitchc·7h ago
> That said, if AI code is _not_ copyrightable, does it automatically go into the public domain? Or is there a third category?

It means that AI-generated code, including proprietary code, is not covered by the existing license mechanisms, including EULAs. Companies will of course handwave over this, but it would not surprise me if this is challenged in court at some point.

0
hxugufjfjf·7h ago
Maybe the Do What The F*ck You Want To Public License (WTFPL). Its a super-permissive open-source license that acts almost like a public domain dedication.
0
TheOtherHobbes·7h ago
The distinction is between creation and distribution.

Even if you accept that AI code isn't protected by copyright, code isn't public domain unless it's released. The legal issue would likely revolve around what 'released' means.

Code that runs on private servers isn't technically released. Code in web pages technically is.

But ultimately legal decisions decide to a large extent on who has the most money.

The frontier labs may eventually try to argue that they own the IP of all code, content, and inventions generated using their technology. That argument may or may not stand, and it's more likely to be over patent rights than plain copyright.

But there's a fair chance they'll try it once AI is unavoidably embedded in everyone's workflow.

0
redwall_hp·3h ago
Copyright applies (or doesn't) from the moment of creation, as per international treaty in all Berne Convention signatory countries. Distribution is irrelevant. Otherwise the law wouldn't serve the original intent of preventing authors from being ripped off by publishers.

"Public domain" is a synonym for "copyright does not apply or no longer applies." There's nothing special about it beyond that.

Something that is public domain can also infringe upon a protected work and you can be liable for that. And something infringing, used internally by a business, could also come out in a legal discovery process.

0
g-b-r·7h ago
Code without a copyright can generally be distributed and put in the pubic domain, and I wouldn't bet that the few existing court decisions won't be reversed.

But, the license reflects the rights that a copyright owner granted you; I don't see why laundering copyrighted material with training a model should remove the limitations on the material's usage that the author established.

So, while it makes sense to not be able to attach whatever copyright you want to LLM output, if it's the result of training on copyrighted code, the extent to which you're allowed to use it should be the intersection of the rights granted to you by that code's copyright owners.

It should thus carry a license that reflects the intersection of those rights.

What makes it much worse is that the model trainers don't tell you what specific sources determined the output, so you should probably just assume that it's the result of its whole training, and the only rights you have are those granted you by all the authors of the training material; and, this in all likelihood means no rights, for most LLM models in use now.

0
mr_toad·6h ago
LLMs are mostly trained on works that are not open source, by your own argument any LLM generated work couldn’t be open sourced.
0
g-b-r·6h ago
Couldn't be used at all, rather
0
simoncion·8h ago
> It's important to remember that the purpose of the license is to allow the copyright owner to exercise their rights in a manner they deem fit.

Just as it's important to remember that the purpose of regulations like copyright and patents are to promote the progress of science and the useful arts.

We're not locked in to a suicide pact with what the Founding Fathers decided back in the late 1700s. Congress can decide that all LLM-generated code is automatically covered by -say- the GPLv3, even if no one holds the copyright to it.

If the absence of a particular person to hold the copyright twists your knickers, you could go a bit further and -say- decide that the people of the United States collectively hold the copyright to all LLM-generated code that's generated by US people or entities, and that relicensing of a particular piece of LLM-generated code must be agreed to by every US citizen that was alive when that code was created.

0
glitchc·7h ago
> We're not locked in to a suicide pact with what the Founding Fathers decided back in the late 1700s. Congress can decide that all LLM-generated code is automatically covered by -say- the GPLv3, even if no one holds the copyright to it.

I have no issue with changing the law. My comment is about the article and the author's misapplication of the law.

0
simoncion·6h ago
> My comment is about the article and the author's misapplication of the law.

What parts of the six-paragraph Fine Article misapply the law? Be sure to quote and include the context surrounding the quoted section(s).

0
stale2002·5h ago
I think it would be the part about how courts have a fairly strong status quo bias and aren't going to remake the entire economy overnight with completely unenforceable provisions.

What are they going to do confiscate the world's gamer GPUs that can run this stuff all locally?

It's completely ridiculous.

0
Joker_vD·7h ago
> you could go a bit further and -say- decide that the people of the United States collectively hold the copyright

Was going to comment something similar; we do have concepts of "dominium eminens", and "escheated property", and "public domain", and many other similar ideas. If the people (through their representatives) decide on a certain property regime, it absolutely can be implemented even if it's different from the status quo.

0
wiremine·8h ago
> even if no one company or person owns it.

Maybe, but I don't think there's any legal president for that, is there? Much of existing law is based on centuries of private property rights. I don't see any government undoing that... or at least, not quickly.

0
simoncion·8h ago
> Maybe, but I don't think there's any legal president for that, is there?

  We're not locked in to a suicide pact with what the Founding Fathers decided back in the late 1700s. Congress can decide that all LLM-generated code is automatically covered by -say- the GPLv3, even if no one holds the copyright to it.
Article I, Section 8 of the Constitution and its various subsections pretty clearly gives Congress this power.
0
Joker_vD·7h ago
The IP laws are a) not really modelled after material property rights; b) are quite much more recent than "centuries"; c) there is legal precedent for that, see e.g. the airplane-related patents being pooled in 1917 by the demand from the US government.
0
mbgerring·4h ago
This is the way
0
stale2002·5h ago
Nobody is going to interpret the laws that way. That's completely stupid.

Everyone is using AI generated code in all sorts of ways, and the idea that the courts are just going to say "nuh-uh. All existing use of AI code is illegal" is completely ridiculous.

Courts don't upend entire economies like that. This would be the remaking of the entire software industry overnight and it would be completely unenforceable.

I'm certainly not going to listen to any such ruling and will continue making code locally. As will basically every other software engineer in the industry.

0
simoncion·5h ago
> Nobody is going to interpret the laws that way. That's completely stupid.

You're absolutely right! And a little more than two hours before you posted this comment, I posted this [0] one! Make sure to read the parent comment for the context... it's pretty important.

[0] <https://news.ycombinator.com/item?id=50507914>

0
stale2002·5h ago
Gotcha, not really related to my point that the courts aren't going to completely remake the entire software industry over night and neither is congress.

That's absurd. It is a fantasy. Neither lawmakers or anyone else is going to remake the software economy.

0
simoncion·3h ago
> Gotcha, not really related to my point that...

Your point was entirely off topic, and the retcon of your intention is weaksauce.

> It is a fantasy.

Reread the title of TFA.

> Neither lawmakers or anyone else is going to remake the software economy.

LLM-generated code is -when measured in regulatory timescales- brand new and is -according to boosters- "remak[ing] the software economy". The time to make new law and regulation to make it prosocial, rather than profoundly antisocial is _now_.

0
semiquaver·7h ago
I really have to wonder what context this person is operating in. The vast majority of software written in the world is proprietary, written for a company to aid some business process, and ridiculously specific to its use case, and freely intermixed with proprietary business information.

“Open sourcing” is not just a legal state, it’s an affirmative act that takes work to prepare the codebase for public consumption. Saying that all the weird business code in the world should be involuntarily publicized is an incoherent concept. What precisely would OP expect to happen in a world that adhered to their opinion? Because to be clear, 90%+ of all code being written today is LLM-generated, so OP suggests that companies should no longer have any privacy wrt their internal software, all LLM code just gets yeeted into the public. How could that possibly work?

Given how easy it is to write software now, this perspective seems to overindex on a past state of the world that no longer applies.

0
Maxion·6h ago
> Given how easy it is to write software now, this perspective seems to overindex on a past state of the world that no longer applies.

This opinion by OP also assumes that code itself has some form of value. With the ways LLM work now code by itself is nigh-on worthless and as the models are becoming more capable, the value of code keeps going down.

0
user43928·1h ago
Agree. I'm not worried about my last 8 months of work, or my employer's for that matter, going into the public domain.

It's obviously not going to happen.

0
g-b-r·7h ago
Companies wilfully betted on AI, usually aware that there were risks.

Maybe companies should just stop using LLMs for 90% of the code?

0
semiquaver·3h ago
The risks you want there to be aren’t real. Neither is the fantasy that AI will go away.
0
g-b-r·3h ago
If you say so (I'm sure that AI won't go away, far less sure that the current crap won't).
0
x3haloed·7h ago
Kind of a silly position. I mean, if it's you're code verbatim, then yeah, it's a violation. Shouldn't be allowed. But like... it saw open-source code and therefore all output must be open source? I don't know about that.

Here's where I can get on board. I think people should be putting AI to work to improve everyone's lives for free. Your version of "should" is like "we should make this a law." My version of "should" is more like a moral recommendation -- we should be using as much of our subscriptions as possible to benefit everyone instead of enriching a handful of people. I view it as a natural extension of the spirit of free software to begin with. Software should be free (as in freedom), and we finally have enough development power to make real headway on that.

0
anself·8h ago
Strongly disagree.

Just because AI was involved doesn’t mean significant human thought and effort didn’t go into the production of those artifacts.

My agentic software development loop requires heavy involvement of real-world feedback. The ratio is huge. 20 mins of agentic software coding against hours or days of testing, trying it out, real-world contact, thinking about what feature is next or what tweak would make it more usable, smooth, intuitive. That feedback is baked into the product. Forcing it to be open source steals all that time and effort for me, and takes from me the decision as to how and whether to open-source it. So you mean the code that I’ve agentically written to help manage my private financial affairs (on local models, for privacy) or my kids’ complicated school schedules must now be open sourced?

My manners begin to fail, so I’ll stop here.

0
PeterisP·7h ago
US legal practice (e.g. https://en.wikipedia.org/wiki/Feist_Publications,_Inc._v._Ru....) is that "sweat of the brow" is irrelevant and the fact that something took significant effort doesn't mean that the product deserves copyright protection.

Furthermore, copyright protects only the expression, not the idea - reimplementing the same idea with "different words" has always been permitted under copyright law (patents may protect the idea, though).

Of course, the fact that it would lack copyright protection it doesn't implythat you have to do something with the code for management of your kid's schedules, just that if someone else somehow legally gets hold of it, the law won't prevent them from making and distributing copies or derived work.

0
jdiff·8h ago
Thought production isn't copyrightable. Processes and procedures aren't copyrightable. Might get you a patent.
0
ginsider_oaks·7h ago
> Forcing it to be open source steals all that time and effort for me

You are the one "stealing" the time and effort of thousands of people who have written free and open source software. If the ratio really is so huge, then just cut out the small part that does all of the stealing.

> So you mean the code that I’ve agentically written to help manage my private financial affairs (on local models, for privacy) or my kids’ complicated school schedules must now be open sourced?

Open source licensing is only relevant when distributing the software, if you just use it privately it has no effect. You are not forced to make it public.

0
mirsadm·5h ago
That would like require learning to code. No one wants to do that anymore.
0
lelanthran·5h ago
> My agentic software development loop requires heavy involvement of real-world feedback. The ratio is huge. 20 mins of agentic software coding against hours or days of testing, trying it out, real-world contact, thinking about what feature is next or what tweak would make it more usable, smooth, intuitive. That feedback is baked into the product. Forcing it to be open source steals all that time and effort for me, and takes from me the decision as to how and whether to open-source it.

It doesn't need to be open-sourced: if it's legal to take an LLM that was trained on every other product in existence and use that LLM to produce clones of all those products, then it's legal to do the same to your product to.

So, go ahead and release. There'll be an open-source clone in no time if the product is any good at all.

In fact, if there are no clones, take that as direct feedback that you wasted your money on tokens - your product is so bad no one even wants it for the few dollars it will take to clone it.

0
bar8uyr·3h ago
Using vibe coding as a crutch will likely not help you improve at writing code. If it has, you may find your code can be replaced by someone else vibe coding. Calling it an "agentic loop" may not change that.
0
rsoto2·3h ago
Yeah, I worked hard to build a factory in a foreign country to make counterfeit handbags, why *shouldn't* i be able to profit off of them.

Nobody is asking you to open source your home app they are talking about corporations stealing code via LLM laundering for profit.

0
otabdeveloper4·8h ago
> Just because AI was involved doesn’t mean significant human thought and effort didn’t go into the production of those artifacts.

Doesn't mean it did either. (99% of the time it didn't tho. Nobody reaches to a vibe code machine because they care or want to dedicate effort.)

0
tim-projects·7h ago
I have some projects I've been working on over a year. Sitting down working making design decisions. Research. Using ai doesn't mean I did no work. It doesn't matter if the ai was trained. I still did work.

I don't understand this view point. Using ai and making high quality stuff is not pushing a button. Its hard. It takes real effort and diligence. And you have to pay actual money out of pocket to deliver it.

0
ozozozd·6h ago
It also takes effort to copy something, right? Say I made a copy of a book by hand, does my effort make it any less of a copy?

Say I am stealing stuff that’s really heavy, can I demand I earned it because it was a lot of effort?

I am not trying to be facetious. I also question property rights because someone overpowered someone else to make that property theirs at some point in time. Whose effort is earned, and whose is robbery?

As far as I can tell we picked an arbitrary time after which all forced acquisition is considered robbery.

That being said, open source is for everyone to benefit. Violating an open source license is sort of a crime against everyone, not just the creator.

0
lelanthran·5h ago
> I have some projects I've been working on over a year. Sitting down working making design decisions. Research. Using ai doesn't mean I did no work. It doesn't matter if the ai was trained. I still did work.

> I don't understand this view point. Using ai and making high quality stuff is not pushing a button. Its hard. It takes real effort and diligence. And you have to pay actual money out of pocket to deliver it.

You seem to believe that because you put months of work into this, you should have rights over the result, right?

But once you release, anyone can spend a few dozens of dollars and replicate it. Do they get the same rights over the clone that you got over your product, even though you potentially spent months coming up with it and they cloned it in a day?

There's really only two answers to this - either you agree that they should get the same rights over the clone that you got over your product, or you don't agree that they should have the same rights over their clone that you have over your product.

In either case, you are not better off:

1. If they get the same rights over the clone that you get on your product, you can't really stop people cloning your stuff, adding a feature or two, removing another feature or two, and then release something better than you for 1/10th of the cost of production.

2. If they don't get the same rights as you did because they used your product as training material for a clone, then wtf do you get any rights at all, after using millions of other people's products as training material?

I don't see any scenario where there is an outcome beneficial to you. Do you?

0
_aavaa_·4h ago
This is a false dichotomy. In this example the person you’re replying to isn’t wholesale copying someone’s work.
0
lelanthran·4h ago
> This is a false dichotomy. In this example the person you’re replying to isn’t wholesale copying someone’s work.

So? I said "clone", but clarified that features may be added to the clone to make a better product.

Taking his product $FOO, adding it to my training set/context/RAG/LORA and/or tool call to explore his product in detail before I tell my LLM to produce $FOO+1 is exactly the same thing that he did.

0
bartnp·4h ago
3. Others are free to use the produced material as educational/training material if they can gain access to the code to then use it for their inspiration. But they are not allowed to outright duplicate it.

This seems the obvious answer? No? The exact clone at 1/10 the cost is just really unlikely imo, especially if it's closed source. And with open source this was already a problem anyway, you could always fork, but hey, turns out most people just stick to the original anyway

0
lelanthran·4h ago
> No? The exact clone at 1/10 the cost is just really unlikely imo, especially if it's closed source.

The availability of the source is irrelevant; I posed this question with the understanding that I could simply add your product, whether closed or not, to my LLM, and get it to produce a better product without all that pesky effort and token usage you used.

Release your product. See how quickly it gets "cloned". i.e. copied and reproduced without anyone needing the source code.

All that effort of yours... just to result in someone else producing the same product in a different language, but with all your product decisions already baked in for nowhere near what it cost you.

> And with open source this was already a problem anyway, you could always fork

Sure, and people famously did so, but they could not clone original closed-source products without a lot of effort, often taking months to years. Now, the result of you spending years on perfecting a product means that a clone can arise in a matter of days.

0
xcf_seetan·4h ago
What if I am creating something that does not exists yet using AI? So i did the research, layout the architecture, decide on the implementation details, write the documentation and the rules AI has to use to create the software, because I am only one person and I can't physically write all that code on my own on a timely basis (it would take me years to write all that code, test it and debug it until i am satisfied)? I am creating something new, not copying anything, so where I stand? Can i publish my code, say with MIT licensing, and retain the copyright?
0
or1gaminal·6h ago
Absolutely. There is so much variability in the code that can be generated for a particular purpose. I do hundreds of iterations before my LLM-generated code reaches the desired shape. The built harness, the written AGENTS.md and the iterations affects what is written greatly.

And at least of the level of feeling the answer is that it is my code.

0
g-b-r·7h ago
It's unfortunate that you had to pay actual money and effort, but there was also actual money and effort involved in the projects on which the LLM output is based.
0
tim-projects·6h ago
When a carpenter crafts a chair, the tree first had to make effort to generate the wood.

Because chairs are in effect an open source design, does that mean carpenters shouldn't get paid?

Its nonsensical.

I think that people who believe these arguments have bought into ai hype too much and genuinely are fooled into thinking pressing a button to spit out boiler plate code somehow equates to a genuine saleable product. It doesn't and it won't. And that stuff shouldn't be open source because it's a waste of everyone's time to consume.

0
dmitrygr·6h ago
The difference between the input to the carpenter and the output of the carpenter is significantly greater than the input from you to the LLM, and the output of LLM. Majority of the alterations to the wood by the carpenter produced the chair. The words you typed into an LLM are less than 0.000000001% of the total inputs that it took to create its output (because training data counts, and it’s a lot bigger than anything you may have typed)
0
g-b-r·6h ago
Yes, what you wrote is definitely nonsensical.

Chairs can be covered by design protection.

And however hard a carpenter might work on a chair, if he copies a protected design he can't sell the chair.

However much work you might add on top of other people's copyrighted code, you don't get to sell it and use it however you want if you don't have the permission of the original authors.

0
lelanthran·5h ago
> However much work you might add on top of other people's copyrighted code, you don't get to sell it and use it however you want if you don't have the permission of the original authors.

It gets even more hilarious when you realise that the GP (tim) does not consider that, if you make the output of AI that was trained on copyrighted material legal to re-copyright, then it applies to him as well - all that hard work he did can be used chucked into an LLM either as context, or via LoRA, or even via simple explorative usage via tool calls, and then regenerated from that LLM, taking away whatever copyright the product had by generating a competing clone.

https://news.ycombinator.com/item?id=50509402

0
emacdona·1h ago
Well, standard disclaimer: I am not a lawyer.

That being said...

I can kind of buy the argument that we humans were trained on open source code when we read it... and then many of us went on to write closed source code. I would also, say, however, that any laws we had in place never expected a single entity to "read" the entire output of human civilization. But that's not really the point I want to make.

The point I really want to make is this:

IF we accept that what the AI companies did when they consumed the entire internet for training was legal (or even ethical)...

Then I fail to see how training other models via distillation of their models is ANYTHING BUT legal (or even ethical)...

0
mirmor23·6h ago
This article is mostly ragebait, and it's claim is hollow when looked thru the lens of computing history.

Let us skip the whole Bell Labs C/C++/unix/sh/ksh -> BSD csh/vi/bsd/tcp/ip, and other proprietary unix variants/utils and focus instead on today's landscape.

There are two axes we can partition the whole ecosystem. Free/OpenBSD, Clang/LLVM, musl, busybox, X11/Xlib, sqlite, postgresql, nginx, Python, Go, Rust, many embedded rtos, vendor sdk, utilities etc.. and then there is Linux/GCC/glibc, etc. Along the licensing spectrum there is GPL and then the rest of the permissive licenses MIT/BSD/Apache etc where source code publishing is not a requirement.

These permissive license offerings are more prevalent compared to the GPL offerings and as such one can gain all the knowledge needed to build a fully functional system. LLM could memorize, or copy the pattern, so even though it is not a clean room, courts seem to be ok as long as things are not spit out verbatim.

Add to this, the fact the there are tons of research articles, user manuals, RFCs, so one can learn and understand how to build the needed reusable component if small and give their own prompt (and bugs) and and let LLM do as user described.

In the US, purely AI-generated code isn’t copyrightable, but it also doesn’t mean you have to publish it. Also, putting an open-source license on code nobody holds a copyright to has no legal effect. So, ethically or otherwise, there is no strong and convincing argument that open sourcing is the right thing to do.

What 'creators' could do is to compile a list of all the frequently used permissive licenses that are applicable to their system and just make it part of their software license attribution.

[1] https://redmonk.com/sogrady/2026/03/25/open-source-licensing...

[2] https://arxiv.org/pdf/2211.15533

0
dava96·8h ago
Whilst the first training sets would have used Stack overflow and open source projects.

I have my suspicions that the AI companies have been using reverse engineering to gain more training data.

Ethics don't really concern these companies so...

When you have essentially infinite compute and are constantly looking for more training data. To me the next logical step is to start reversing compiled programs.

It could also explain why these AI models are so good at it too...

How do you copyright stolen material?

0
singpolyma3·7h ago
Reverse engineering is legal so I'm not sure what you're trying to imply here
0
novok·5h ago
The other take is AI code generation machines are directed by human inputs and will never start without that human input, so therefore it was created by a human.

Much like how a photographer 'just' clicks a button on a machine and through a very complicated process it creates media, so does the human who makes media via an AI machine.

I think because of how corporate power works and how much money is at stake this will end up as the end result. Laws are made up by humans and a majority vote in a parliament can set this to be so, previous rulings be damned. And most treaties never really ever think about how media was made because this was never an question until recently. Copyright are not constitutional rights in most counties.

0
asdfasgasdgasdg·1h ago
> The second reason is legal: purely AI-generated code is not written by a human, meaning it should not be copyrightable.

Even if this were correct, a piece of code not being copyrighted does not automatically confer rights onto third parties to have it conveyed to them. As a matter of practice, what more than any other factor keeps proprietary code proprietary is not the legal shield of copyright but the difficulty of getting companies to convey that code to you.

0
saikiran91·4h ago
I don’t know about one-shot-prompt toy apps, but when it comes to building real stuff with AI, it still takes significant effort from devs to get the results we need. There’s a difference, right? “Give me the best result” vs “Give me the result I want”. Most serious projects, at least from what I know, run on the “Give me the result I want” principle, and that’s where the dev effort goes. I know the future is heading toward “Give me the best result with no mistakes”, but that changes how much effort it takes, not who owns the output. Even if the code is generated by AI, humans own the result, because they worked with the machine to get the result they wanted. Same was true before AI: humans worked with the machine to get the result they wanted. It’s just way faster now. Using a machine to build something has never meant you don’t own what you built.
0
tonymet·4h ago
Nearly every anti-AI argument seems to be someone who’s limited experience is one-shotting.

Most of my AI experience takes quite a lot of supervision just to get the model in the context to be able to make changes effectively.

0
Sparkyte·4h ago
The problem with this is the assumption that models used are not a factor of the code already written by the user and ingested as context to determine the future state of the code according to the specifications.

All purely AI-generated code should be Open Source.

I've got plenty of technical designs that should not be Open Source and follow distinct design decisions AI can work with, not necessary know how to deal without context. If AI is additive it is just adding on the existing design not changing it.

Also I would dislike AI rot following into the codebase of Open Source projects making them unmaintainable.

End of the day question is if AI is driving you or are you driving the AI?

0
sicher·8h ago
I mostly agree and have pretty much every one of my projects open source. One in particular, though, is not. And that is because the design of it is heavily domain specific and is built on loads and loads of experience. Not sure how to reconcile that...
0
regularfry·8h ago
There's a difference between something being released as open source and something that never gets released at all. Open source or not only becomes relevant when making the binary available to other people.
0
sicher·7h ago
Yes, but the one project I'm excluding from open source may be one I sell to companies - we'll see.
0
glenpierce·8h ago
I’m confused. Is this article meant to be a scholarly legal argument (albeit, by a novice)? Or a normative statement about what the author would prefer?

It certainly doesn’t seem to be about practical solutions. It doesn’t even touch on the ramifications of its suggestions. I’m pretty sure that it would be a massive boon to data center operators and cause most of the readers of hacker news to lose our livelihoods.

0
hintymad·3h ago
> All AI-generated code should be Open Source

What problem does this solve, though? If anything, wouldn't this help empower the frontier labs, in particular Anthropic? With everything open sourced, the labs get the knowledge of each company for free, right? Thank about how IBM owned the computation with mainframes until personal computers and workstations became powerful enough.

If we want freedom as the open source movement advocates, we'd need focus on make building open models accessible to millions of engineers and companies. Given the sheer investment required to build the frontier models, though, I'm not sure there's a way unless we have breakthroughs in the AI algorithms or hardware, so that individual engineers or small companies can build competing models.

0
userbinator·1h ago
AI has already "opened" the source of lots of software anyway, so IMHO this is a moot point. However, what AI is making more people realise is that everything is a derivative work.
0
WhiteNoiz3·6h ago
Personally I am less concerned with all AI generated code being open source. We will very quickly be drowning in a sea of code, and who even wants to see it if it's easily re-created by AI? Eventually everyone will be using AI and I don't think it's feasible or makes sense for all code to be open source. Personally I _would_ like to see all models trained on public data become open-weights. To me it seems unfair to let a few companies lock up all the value of the collective works of civilization without providing something back for free.
0
trentnix·5h ago
> purely AI-generated code is not written by a human, meaning it should not be copyrightable

There's a lot of room in between code that is closed-source and code that is "copyrightable".

Should code crafted via cut-and-paste from StackOverflow be open source? How about newsgroups before that? What about apps that use "no code" tooling? Is there a threshold for the use of external libraries and packages that cross the threshold for "that code was simply scraped from the internet and used without permission or proper attribution"? How about code I wrote while using auto-complete a lot?

I appreciate the author's sentiment and recognize that LLMs have aggregated the work of many, many people. I'm guessing their work wasn't wholly original either, like so much of the code I wrote before the age of LLMs. Whether those people whose work has been aggregated to train LLMs own that work - those ideas, those thoughts, those strategies - is exceedingly complex and (I speculate) largely won't hold up to real scrutiny.

0
mentalgear·4h ago
Actually: All AI-generated _Output_ should be Open Source (not just code)

It's the (open) knowledge of humanity, and the thin layer above is only a minor fraction (0.1 if at all) of the 'product', which should not be allowed to be privatized by anyone - let alone destroying originals (like books) to own 'it' !

0
Kevin_Flynn·7h ago
I can understand the case being made on "purely AI generated code."

But thats also why its buggy.

As soon as I start refining it, and commanding the AI to fix the mistakes, this argument starts to lose ground fast.

The tech community has adopted open standards for as long as I've been alive, and the AI is using these open standards as much as it is actual code.

Perhaps an open standard to prevent code from being used as training ( it probably already exists, i dunno ).

Methods to detect it ( I've seen that showing up recently. Using AI to detect AI. )

Some sanctioned actionable penalty under provable circumstance as the resolution ?

You can see why the copyright office wants to treat this as any other thing.

The details of proving claims get complicated quickly.

0
antonvs·7h ago
> As soon as I start refining it, and commanding the AI to fix the mistakes, this argument starts to lose ground fast.

Definitely. And if the result isn’t essentially a clone of some existing copyrighted work, there really is no viable argument under current copyright law.

0
osigurdson·3h ago
Open sourcing all AI generated code isn't practical and doesn't really even address the issue. An AI model is never going to be able to properly provide attribution or determine what code used what license anyway. Instead, GPL code should be removed from AI training sets going forward by default. Essentially, anyone who does not want AI to train on their code / work of any sort should be able to opt out.

While this will not impact current models, future models would eventually be free of it.

0
coffeefirst·7h ago
He’s not wrong, but I would argue a better model is the Derivative Work. This already exists in copyright law.

If I write a space opera that is a LOT like Star Wars there are major legal implications to this. If I write an original work of science fiction involving a weapon that destroys planets as an allegory to the history of the atomic bomb, I’m probably in the clear even if one of the characters is a furry alien with a laser sword.

The challenge now is how to apply that framework when nobody can see the original or resulting code.

0
chaosharmonic·7h ago
Derivative works still have their own copyrights, though. For instance, Kouta Hirano might own Hellsing, but that doesn't automatically mean that he owns enthusiastic walks through the woods, "bitch I eat people," or High School of the Dead.
0
drsalt·3h ago
AI companies should be required to pay repartitions/royalties to the originators of their data. But people aren't ready to talk about this yet.
0
brisky·3h ago
First step for this has to be requirement for AI companies to provide lists of data sources that their models were trained on in addition to each model release. Then will come the lawsuits and then will come the royalties model.
0
anat0la·3h ago
This is the government responsability but they are busy harassing ordinary people and keeping the wealthy people unchecked. while we all suffer
0
Buttons840·3h ago
In the process of trying to track down who had the original idea we might discover nobody actually has original ideas. That would be awkward for IP law.
0
al_borland·8h ago
What's wrong with this website? When I open it in Safari on macOS, it consumes all my RAM. 20GB to start. I got a system warning to force quit some apps to maintain stability (which was a first in 5 years). When I closed other non-Safari apps, Safari usage then went up to 30GB.

Quitting and relaunching Safari, then returning to the site, produces the same results. The process for this website is what is using the RAM, it persists after closing the tab, and needs to be force quit.

0
DustinBrett·7h ago
Do we still care if its open or closed? Isn't reverse engineering basically solved anyway?
0
monneyboi·2h ago
I wrote up my thoughts about this a while ago here: https://resolve.works/articles/the-tool-is-not-the-author/
0
westoque·4h ago
i think the author presents weak arguments:

> AI models were trained on a massive corpus of Open Source

using open source shouldn't require the final product to be open source. that's pretty much most of proprietary software. most of it is built using open source code but the final product is due to a combination of thoughtful product and software knowledge and must be private since it's really the value you produce.

> purely AI-generated code is not written by a human, meaning it should not be copyrightable.

you can say AI generated code is based on human code, and that human code is open source. depending on the license (likely MIT) its technically legal to choose the generated code to be private.

> To make things fair across all jurisdictions, and to honor the open ecosystem that made AI possible, developers and companies should explicitly release AI-generated code under Open Source licenses or dedicate it to the public domain using the Creative Commons Zero (CC0) license.

i somewhat agree here. AI generated code should include the original license of its authors unless it's purely AI generated using original variable names.

i previously released AI generated products/software as open source but come to think of it, in the world of abundant software, i believe there's more reason to actually privatize your code vs putting it out in the public. especially for paid software products.

0
Shugyousha·4h ago
Matthew Butterick (of https://typographyforlawyers.com/ fame) is involved in a class action suit about LLM being trained on copyrighted material. See https://llmlitigation.com/.

Maybe we should get some high-profile software devs on that as well ...

0
pizzly·5h ago
Two points I think making AI-generated code opensource Point 1: AI can (or will be able to) reverse engineer all code from how the program works. Thus, there seems to be little point in opensourcing AI generated code because it can just be reversed engineered

Point 2: Correct me if wrong but can AI generated code actually be copyrighted? If someone just keeps prompting but not touching the code would this still be considered pure AI generated work? If this is the case then point 1 and point 2 basically means AI generated code is already open source in effect if not in name.

0
vishnuharidas·2h ago
Not necessarily IMO. It's not the code, but the IP is what matters. The IP holder should have full rights to make it open or not. Using a tool that leverages open knowledge doesn't prevent someone from building closed solutions.
0
mlinksva·4h ago
Good link in the post https://commons.wikimedia.org/wiki/Commons:AI-generated_medi... happy but not surprised that Wikimedia Commons has a good approach and I lovely template name as part of it: PD-algorithm
0
gwynforthewyn·5h ago
A lot of y'all here are missing something important: open source code is an application of copyright law, and it has already become clear in some countries that the output of an LLM is not coverable under copyright (i.e. it is public domain).

In the UK, the output of an LLM can come under copyright and therefore the author can license it as an open source product, but in the US it can't be done.

0
maxdo·5h ago
Alternative view on the problem. if code is really a solved problem , open source becomes a security problem? with all the attacks happening mostly on open source projects, it is not farfetched that a secure bank system will be ai generated with core libraries implementing everything from scratch, including runtime. Big business, big problems.

This idea of no libraries/external dependencies sounds extreme, until you got hacked by a tiny library for http calls and that cost you billions of losses.

0
mococa·1h ago
To be honest, AI generated code must never have a license or a very permissive license
0