- cross-posted to:
- hackernews@lemmy.bestiver.se
- programming@programming.dev
- cross-posted to:
- hackernews@lemmy.bestiver.se
- programming@programming.dev
cross-posted from: https://feddit.online/c/tech@feddit.online/p/2115447/all-ai-generated-code-should-be-open-source
So if a programmer googles which parameters a certain function accepts, that would mean Open Source? Better definition is needed.
Maybe going against the wind here, but this doesn’t make sense to me for generated ideas and code to be open source. If I have a unique recipe that made something valuable, I should be able to have that power of decision over my idea. What I think would be a must for this to be realistic is that all LLMs that trained on public data be, by law, public domain. So a company would still own how they did the training so there still would be something where that expertise is rewarded.
I agree, it’s more about the economics of AI use itself rather than the end product that’s concerning to me.
I agree with this.
I’m not a programmer or anything, but it seems to me that AI may help in this regard (for example, the PS5 hacks allowing people to play old titles that Sony can’t be bothered with)
The mystery LLM machine has been trained on all open source code (also known as stealing in the era of the past), and also on corporate code (also known as fine-print-legalised incidental training). Essentially, if you have freeish access to big llms, it is all free (ish) and open source now, especially code from companies that went gun-ho into AI, smth smth creative suite…
If it’s all free and easy then just go and get your own AI agent to do your programming. Not so easy now.
This is like saying books shouldn’t cost money because we have printing press.
All code should be open source, period.
Everything should be free. amirite?
What do you do for a living? What ever it is you produce, I want it for free.
The people I’m providing my service to do indeed get this service for free. But what has that to do with open source?
I already wrote you don’t have to make money with code. You can offer support and be payed for that, or you offer to fulfill feature requests.
Or you can bind a license to your code if you absolutely want to squeeze something out of it. Just because a code is open doesn’t mean that you have to publish it under a license that allows for free use. You know the funniest part of this? If every code was open, it would also be very easy to proof that someone used your code or parts of it illegally (according to your license).
I guess the key is that ‘open source’ generally means ‘free use’ versus ‘source available’ which at least in the old days used to be a supremely common arrangement for commercial software.
I will also agree that every other IP pretty much is effectively ‘source available’, you see a book, there’s no hidden secret form you need to construct that work. Nothing inherently stops you from reproducing a modified copy. Sure there are more convenient assets (having distinct tracks that are used to compose a song versus only the post-mix recording), but generally other IP has to decide how to deal with remix, sampling, etc, without some fundamental barrier like ‘it got compiled into machine code which is hard to modify’.
Hm, yeah, I guess the thought of ‘open source = free use’ is the issue that prevents many from entertaining the idea. Makes sense.
Especially now - let’s say you GPL some code - I have an LLM read it and write equivalent code but, so that each line is different.
Before you could put it out there, and ask “use this but give back.” Now it’s just “pretty please don’t steal from me. Just kidding, you already did.”
Well, I didn’t say anything about you immediately going public before you have something concrete. That’s your decision. Opening up the source should be one of the final goals, though.
However, I feel like this argument is mood, because if it was that fragile, open source and companies that make money with it would not exist at all. Might be because someone ‘stealing’ code does not necessarily introduce new things, the original authors though might. Which makes users stick to the original, rather than the copycat, because the copycat does not move unless the original does.
Yeah. I’m not arguing with you.
I’m just saying the landscape of [L]GPL has changed. Even a MIT license - the implicit gravity of a code-repository where the experts contribute - it has … changed.
Whether all of this is for the better or for the worse (I mean, being able to hack your devices is pretty cool, and I personally can’t wait for the iPhones to be completely hacked) – we’ll find out.
I think you just invented socialism. Isn’t it great!
Well I understand the justification for human produced code not being open source. It could have taken them, hundreds of hours, depending on the program, to develop. They need to be compensated for that time.
But AI, it can recreate Photoshop in a day and it’s not even as if anyone had to sit and stare at the screen the entire time. That being said it’s not resourceless, somebody has to pay for the power it took to generate so even AI produced code probably should cost something but not anywhere near as much as human produced code electricity is a lot cheaper than human time.
Let me rephrase that. In an ideal world we would open every code. You still can make money with support or feature requests and so on. It just would make everything more resilient and easier to fix. Having said that, we’re not living in an ideal world, so this will remain a philosophical debate forevermore.
As for the AI part, if I’m playing advocate of the devil here, energy costs are one thing. The other is, forging the prompt and iterating on it still needs humans. There is work in this process that can be worth a compensation.
So, would you say open up the AI code, but allow for the prompting to be “closed source”?
hotter take but i’m right here with ya
Absolutely insane, terrible take.
Code is intellectual property just like anything else. You should be able to make something and then decide who can use it and under which conditions.
And I’m pretty content with my banks code not being open sourced.
Security through obscurity is not security. The algorithm for SHA-256 is publicly available, doesn’t mean that you can use that knowledge to reverse the 1 way cryptographic encryption on it, nor can you brute force it with the computation power of the equivalent of 37 observable universes from big bang to heat death.
Security through obscurity is how you get the eMudhra 4 incident.
Yes yes I know. I was just making a list of the top of my head, you win the technical point that this specific example isn’t great.
There are no examples where obfuscation of code is an inherently good thing (except to hoard knowledge for profit motives). I use Blender, the FOSS 3d software that at this point, easily competes with and even imo, outperforms established proprietary closed source software and has a plugin community so robust that if thers is something in another program that isnt in blender, its very likely someone has made an equivalent as a, often free, sometimes paid plugin that does the thing. Were also at the point where the proprietary software has been taking design cues from Blender and applying them to their own proprietary programs.
I would want to know that my bank is using SHA-256 to encrypt passwords and my information.
In general, I have an issue with the notion of “intellectual property”. There is no reason to restrict what is essentially knowledge. I’d even argue it’s dangerous in many cases. Also, it’s absurd because you are never the only person having the same idea, yet people are being punished for not publishing it first.
As for the bank example - I am not. Opening the code does not mean everyone has suddenly access to your data. Your data is hopefully encrypted well … which you will never know, because your bank keeps everything hidden from you. No possibility for you to make an informed decision, based on your own research or that of experts.
That’s a perfect frame for hackers to operate in. It’s also a perfect frame for companies to cut corners and reduce costs, while getting the maximum amount of money out of you.
In general, I have an issue with the notion of “intellectual property”. There is no reason to restrict what is essentially knowledge.
Yes there is: It creates a financial incentive to create new and better things. This is why patents exist.
See my posts here: https://feddit.org/comment/16096467
In general, I have an issue with the notion of “intellectual property”
As a broad economic policy - providing an economic incentive to create IP is very good. It is was fueled nearly all of the innovation of the last 300 years, probably longer. It also has downsides, but it is a net good. I firmly do not think there would be as much advancements in general if we structured IP in a way that was public / no profit motive.
Opening the code does not mean everyone has suddenly access to your data
No but it means your exact package dependencies are public. Your exact encryption methods are public. The specific ways the data is distributed is public. And so on. This makes it significantly more vulnerable to the various attacks which would lead to leaked data.
No possibility for you to make an informed decision, based on your own research or that of experts
Is there a name for this fallacy? The “Your own research outweighs that of a trained team of professionals” part. Me conducting the research does guarantees intent alignment; it does not guarantee accuracy. I’m wrong about things all the time.
The idea that some random bloke, even a trained software engineer, could look at the massive legacy repositories of these banks and make an “informed decision” is silly. I guess if there was glaring flaws, then maybe. But realistically, most people would just rely on a handful of organizations to audit the repo’s and summarize the security for me except… those sorta already exist. Via things like SOC 2 and what not. Even a trained security specialist spending many months full time dedicated to evaluating the security of the bank (based on their open sourced code) would have a hard time coming up a decision.
For the most part in humanity’s existence, innovations didn’t depend on arbitrary rules of lockdown and monetary incentive. Sure, it helps, I’m not questioning that, but I don’t see why in an open source society this would have to stop. I have a problem, I’ll give you money to solve my problem. When you’re done, open up the code. Why wouldn’t you take the job? Why wouldn’t I still pay for work/help?
As for the second part, there is a lot of security related software which is open source that is still secure, I’m not talking fringe cases, I’m talking industry standards - encryption being on the forefront. Or rather BECAUSE it is open, it is so secure. The method of how you do things, does not enable others to crack the result you have in the end. For instance, we know exactly how some VPN tunnels are being encrypted. Cracking this encryption live though is a whole other story.
Third part. Correct, you can never be 100% certain. Doesn’t mean this is a fallacy. The chance of catching holes in the system rises dramatically with the amount of persons checking said system. You want a fallacy? Then riddle me this. Why would someone trust a company obfuscating its method and selling a ‘trust me bro’ more than multiple (cross-)checks done by thousands of independent individuals all over the globe?
For the most part in humanity’s existence, innovations didn’t depend on arbitrary rules of lockdown and monetary incentive.
For most of human history, we were hunters & gatherers. Even if you mean that start of civilization (~6k years ago), we spent most of that time with significantly less tech. Modern IP legislation started ~300 years ago (different specific times for different nations), which is roughly when technology advancement started to sky rocket
Why would someone trust a company obfuscating its method and selling a ‘trust me bro’
Neither of us think either option is perfect - it’s not about trusting one absolutely. It’s about which one do you trust more, and I trust closed source & highly maintained over open source-audited but also open source vulnerable
I’d argue that sky-rocketing is mainly the result of a self-enforcing/accelerating exponential network effect. The more technology you get the more breakthroughs can be made.
Again you can accelerate that with monetary incentives, yes, but humans are fundamentally driven by curiosity. That will never change. Also again, an open source society wouldn’t have any need to stop monetary incentives.
You’re right. Intellectual property should be done away with entirely.
Mhmm yes, takes that on a surface level seem to benefit everyone but in reality make it much easier for large corporations to steal ideas and leverage them to gain power without granting any profit to the creator.
Wild to believe IP laws work on behalf of the little man. But, yes, you shouldn’t be able to monopolize ideas or strings of words on paper anymore.
It will be, same with pretty much all code. SaaS companies are completely dead as a profitable capitalist business model, no one will pay for software in the future.
I think this is fantastic and I fully support it. Anything made by AI should be free and open source. AI should benefit humanity, not be used to make the rich richer.
I think it’s time we made a new type of copyright, we’ve had copyright and copyleft and Creative Commons, we need something that indicates derivative AI work based on previous stolen content.
First have to get courts to recognize that the training is subject to copyright for a license to apply in the first place, which isn’t looking good.
I fully anticipate that there absolutely will be copyright consequences for AI generated knock-offs of big software, but conveniently the well resourced AI companies won’t be on the hook, just the end users will be left holding the bag. By that point, the license goes back to working as intended already.
There are actually two paths leading to this conclusion: The first is the legal assumption that slop is not copyrightable. But that can be changed if pressure on lawmakers rises. The other part is that those LLMs have most likely also been trained by code under the GPL, making them essentially derived works.
Slop has been copyrightable for as long as I can remember. (Books, movies, etc.)
Yep, but that’s human slop 😃
Pure slop is not copyrightable is likely the case (the couple of cases where they ruled that works with zero human involvement do not grant any human copyright). However, the moment a human tweaks it, that resultant work can qualify for copyright. Consider that Steamboat Willie Mickey is public domain, but despite being derived from that design, Fantasia Mickey is still copyrighted. In the cases where courts ruled the human didn’t have copyright, the human had loudly and proudly proclaimed that there was no human in the creative process. If they claimed they applied some transformative work to it, then that would be copyright eligible. Further, a person is not compelled to divulge the ‘raw’ materials, so even if the slop they tweaked would be public domain, they are under no obligation to show the ‘pre-human’ form of the data.
On the matter of LLM works being derived by virtue of training, unfortunately it is looking like the courts are broadly siding with the AI companies. Not only is it not inherently infringing to train models using material without express permission, it is also the case that if the output is infringing, the user, not the provider, is on the hook for the infringement.
Yeah, but it was also trained on scraped proprietary code and other incompatible copyleft code, so the LLM produced code isn’t copyrightable and can’t really be licensed, open source, or public. It’s undefined.
And if it’s AI-generated, it should have metadata attached that states that fact.
I agree on principle but the majority of code tooling does not reliably track meta data like that, and adding it would be a significant degradation in quality.
IMO, UI that does too, if it is an app with UI, so the user can choose whether or not to use it
This stops short of the logical conclusion if you continue this way of thinking, because it only focuses on code. Every single public comment ever made on the internet (and through the wonders of digitalization also a lot of stuff from before that point in time) by anyone is in there.
These models were trained by the combined cultural output of humanity, which makes them cultural artifacts. A snapshot of humanity which should belong to noone/everyone. Putting them under UNESCO stewardship and licensing them - free personal license, paid corporate usage, reduced prices for NGOs and the like - while using the proceeds to fund the arts should be a given. The licensing should also declare everything generated as being part of creative commons.
Licensing costs could also be used to reduce the amount of corporate slop. Licenses that are aimed at producing material to be disseminated in the public space could be pay-per-view on an exponential scale, making sloppifying the internet an expensive mistake.
Won’t stop spammers/scammers, but that is an problem that already started back when your hosts file was still manually maintained by the Network Information Center at Stanford and had to be sent out by snail mail.
These models were trained by the combined cultural output of humanity, which makes them cultural artifacts.
It’s more like the combined cultural output of humanity arbitrarily selected from, and the results stuck into a blender and liquefied into an unrecognizable slurry.
But what societal values you get as output is statistically what we all would have written (That’s why Grok has to be lobotomized on the regular, because it often disagrees with Musk, especially when it comes to society). Therefor, future generations could always load up a model and get responses that can explain cultural phenomena of the time period. (They could even provide them with current language information, so they wouldn’t have to translate anything themselves). A similar artifact of previous, long gone civilizations would be priceless, even if the output isn’t 100% trustworthy - as are personal accounts; historians are always aware that the writers are biased or do have wrong information themselves, so that’s not really an issue.
These models were trained by the combined cultural output of humanity
Sounds a lot like school.
We all stand on the shoulders of giants. Even my friend, Chaz Jippity.
Yep, it’s pretty much the same process. While the human one takes 2+ decades and involves a lot of physicality, the digital one is quicker but devoid of any personality. For an imperial tech-priest in the year 40k, a GGUF file of a 2025 LLM is practically like an average person explaining things like they were in 2025.
All
AI-generatedcode should be Open Source.And, as people are starting to find out, all code is “open source”, in that it is incredibly easy to decompile anything. How does that t-shirt go? “Everything is open source if you know assembly.”
Well, existing decompiling tools already knew how to reconstitute assembly into higher-level languages, and LLMs can spend the effort to figure out proper variable/function names to make it more readable.
all code is “open source”, in that it is incredibly easy to decompile anything.
Can we stop with this misinfo? The lay person hears decompile and interprets it as de-obfuscated. Deobfuscation is now easier but still not “solved” in the way you are implying.
It’s not misinformation. Entire games are getting re-translated into different languages in a matter of days. PhotoCraft was built in a single day.
It’s spooked indie developers so much that they started pulling their own demos. Copy slop is running rampant on Steam now. Not just copying the idea, but copying the whole damn game, name and all.
Actually i think since digitalization became a thing, everything digital should be Open source.
There is no such thing as property of an intangible, not-scarce thing that information is.
True, it hardly costs anything to distribute a book digitally, compared to selling them in a store. But the work to write the book is the same, and shouldn’t the author be paid for their work?
Commissions are already a thing,plus, there’s a reason why RHEL,SuSe and Canonical win millions over the years, and just doing FOSS.
This is false btw
Imagine if any work made by AI was free use, public domain, uncopyrightable. Bet that would change the equation for corporations a bit.
Make a movie using AI? Anyone can make copies and distribute them freely.
There is a company making a completely AI movie, actually. https://www.cbsnews.com/news/tilly-norwood-ai-generated-actor-feature-film/
I’m pretty sure, at least in the USA, there have been rulings that there must be human authorship for something to have copyright protection. This has happened both in the courts and the US Copyright Office has spoken on the matter as well. If you’re interested in one of the court cases I’d suggest looking up Thaler v. Perlmutter.
There was a stipulation though, any human made parts of a work is copyrightable, so in your scenario of copying a movie would probably still be illegal unless the entire movie was created by gen ai, but it would seem that you could probably cut the gen ai bits and release and reuse them without violating any copyright, but I’m also not a lawyer so…
Your right, but it should apply to the whole work. We need to Stop fucking around and put the brakes on them.
I’m not 100% sure how I feel about that take. Part of me agrees, but also part of me doesn’t. Especially with big projects like a movie it could be easy for one person to slip in a tiny bit of ai generated work and now the whole movie loses its copyright? Seems like a bit too much of an overcorrection to me.
I do strongly agree that we need to hit the brakes hard and really work through a lot of the issues gen ai has brought up.
That gives great incentive for people to stop using it and to heavily restrict its use to only what is needed. If an entire project can be poisoned just like that, larger companies will absolutely hound AI companies to make their output 100% identifiable.
Nuance is the hard part for sure. Im open to ideas.
You could say the same thing in the software world. Right now, and it would have nothing to do with AI.
Literally all it would take is for somebody to slip in some open source code and an entire project is suspect.
This is true for at least ai generated images I have heard.
Make a movie using AI? Anyone can make copies and distribute them freely.
I look forward to that day.
Could you imagine, a worldwide network of people freely sharing movies, music and tv shows for anyone to enjoy.
Maybe eventually we’ll have the technology
Lets call it like river flow or torrent or something.
And NEVER sell them, that should be still a crime
Any time you stick a non-commercial clause to a license, it prohibits a ton of stuff because well, you can’t use any of it if doing so makes you money. Like, want to run a website, or make youtube videos, or stream, and you have ads or donations enabled, or patreon or anything like that? If so, you can’t use NC licensed content.
That’s why one important part of public domain is that nobody is also prevented from selling them, so that the content is actually free for anyone to use.
The thing is that, as AI are created from theft, then their output must never be sold.
I’m of the opinion that IP law is fake and made up, and therefore training AI on stuff isn’t theft, and neither is piracy.
It’s also why everything I’ve ever made is licensed either permissively or copyleft.
That’s only true for big corpos and rich kids, not for the 99% of us.
It’s not okay to steal from an indie, never. And even for small or medium businesses it’s not okay. Piracy is only good and acceptable to the extent that it empowers those who are less advantaged; in other words, it is only justifiable when directed against a millionaire, a corporation, or a criminal (in the latter case, even if that criminal is not wealthy).
Yeah but IP law has never protected the little guy, it’s only ever been used to punch down. I’m not against copyright law, to be clear, just how wide and nebulous IP law is.
The current way that creating something not only gives you the right to restrict the copying of it (which I support) for like 100 years (too long imo) but also the right to:
- decide how it’s viewed, control what people can do with it even if they don’t share it. (what format it’s in, what it can be displayed on, etc)
- decide what people can do with ideas inside of it. (e.g nobody being allowed to add a character from another series into their own.
- restrict the sharing of snippets, like in a review.
is way too much to me, I don’t think that people should be allowed to have so much control over something they put out into the world. To me, you should be able to keep the right to copy your work for, say, 10-20 years and the right to the ideas in it for a shorter time. The fact that superman still isn’t public domain is absurd.
The current system only empowers large companies. When was the last time someone successfully sued a big company for one of the numbered points, versus how many times the opposite has happened?
I don’t like calling piracy theft because it conflates property rights, “basic” copyright and IP rights. It’s obviously still copyright infringement and bad, if it hurts a small developer or artist, which, incidentally, I don’t do. But I pay them because I want to support them, not out of respect for the current state of IP law.
All that to say why I don’t see a problem with training anything on AI, since that falls under point 1 for me. Also, banning it would only make every social media platform add “we reserve the rights to use anything uploaded for AI” to their EULA, and nothing would change, except that open weight models disappear.
Well, I’m with you on that.
Basically and in short: fuck Mickey Mouse
I’m glad you’re able to buy groceries with other income and do that as a hobby. Some people do it for their primary income.
To be clear, I still think pirating non-megacorp productions is bad, I just don’t think we should conflate IP and property rights, and just evaluate whether it’d be a good thing or not to expand IP law to cover AI training.
Which it wouldn’t, since expanding IP law has never been good for the little guy, making companies have to pay to train AI on stuff won’t make them pay the artists, they’ll just pay the record labels and social media platforms, which will all have updated their EULA’s to retroactively give them the right to use everything on there to train AI. None of those dollars will go to artists. The only thing that would change is more money changing hands at the top and removing competition in the AI sphere.
Make a movie using AI? Anyone can make copies and distribute them freely.
A Chinese outfit tried making a full length film using entirely with AI. It flopped hard.
Betcha it would be everywhere freely as being navigated on the high seas, most probably being MST3K’d.
The problem is thst public domain is worse than copyleft.
I’ve been waiting to see some crops come to harvest from this. Big company x say they are building flagship software by machine that cannot create copywritable works.
I mean, the fights in the future courts, but they are setting themselves up for a real contest for ownership in the US at least.
deleted by creator
I have made 3 pretty cool windows programs all released completely open source with generative code assistance. a dimming automation app, a modern experience first media player, and a fully featured weather app tray utility with radar and alerts and everything. at no point did i ever consider anything but open source, and won’t in the future.
This is exactly how it should work. Any code copied from GNU GPL code should legally also be GNU GPL, no ifs, ands or buts.
Oh, your training data set contains GPL code? Guess what, all the output from that model is now GPL. It’s your burden to prove that your training data does not contain any GPL code.
@ free software foundation: your mission, should you chose to accept it ☝🏻
The precedent for copyright work is that the work must be human-authored.
If there is no human author, then it’s not copyright, so no one has the authority to license it, including with copyleft license
It’s all a mess really
It’s nearly impossible to prove a negative. But it would be trivial to ask it for code from a GPL project. If it can provide it, it’s in the model.
It’s nearly impossible to prove a negative.
In a general, open-ended situation, sure. In this case, the model developer has all of the training data that was input for training the model. It should take all of half an hour to determine if there is any GPL code in the training data set.
Hi, I am from OpenAI/Alphabet/Meta, I have checked our training data and it contains no GPL code, or copyrighted or pirated media of any kind, it’s all perfectly totally 100% legit. Thank you for your inquiry, hope this solves the issue and have a good day.
Right, so, then we have a paper trail of fraud. Forcing them to make such a statement in writing is useful.
This guy gets it.
Social media users want to see a person walk into the bad guy’s office and punch them in the face, solving the problem. Real effective steps are usually boring and unremarkable like forcing them to make a statement on record.
Exactly, this is how you win court battles that set legal precedents which establish real-world consequences, not as one-offs but every time somebody crosses that line.
LLMs obviously don’t copy code. That’s not how they work.
What you’re suggesting is sort of the same as claiming that if a developer ever read GPL code, all their code is now GPL code.
Nope, humans don’t generate tokens from a statistical model. We have a fundamentally different way of thinking.
That’s probably too simplistic, but they sure don’t think like humans. But also humans can infect their thought processes in a way that the code they write ends up licensed by some other terms than what they thought it should be.
Like if you have read your employer’s proprietary codebase and write the same code somewhere else, they could sue you for that. Perhaps. IANAL.
Damn that’s a great framing. Will use this talking point going forward.
Collectivize the slop! For teh lulz!
federation test
















