ai delivers killing blow to copyleft

https://malus.sh/ copyleft and things like the gpl have already taken an absolute beating over the years by things such as tivo-ization, software-as-a-service, and huge corporations just blatantly violating licenses knowing the authors don't have the time or resources to take them to court over it. now thanks to ai, anyone can take any open-source software project and use LLMs to automatically rewrite the code in a way that makes it structurally different enough to not be in violation of any copyleft license, and the outputted code becomes effectively public domain due to the fact that machine-generated works cannot legally be copyrighted. tl;dr you can now take any gpl code and turn it into public domain/permissive license code at the push of a button. i guess that about wraps it up for copyleft.

Let's make a version of this that pumps out legally-valid GPL imitations of every single .exe .iso .msi ps4 game and steam game in the known universe. OP needs more replies either way because this is important developments

>>91762 All non-GPL software that exists in the world can now have it's very own personal unauthorized GPL version that any fan can further contribute to after it comes out the oven

How effective would a new license be? I'm thinking clauses that forbid using AI to create "legally distinct" code. Would something like that have a shot? Either way this news is like kicking dead whales down the beach, this timeline is like kicking dead whales down the beach, I wsnt out

>>91763 >Let's make a version of this that pumps out legally-valid GPL imitations of every single .exe .iso .msi ps4 game and steam game in the known universe. copyleft is also a form of copyright, so machine-generated code cannot be copylefted either. anything machine-generated is, according to current laws, uncopyrightable, so any proprietary code reverse engineered by ai would become public domain. but what you are talking about is something that is already beginning to happen and will possibly become a huge soykafstorm in the near future. china's entire industrial economy for the past years has been hyperfocused on reverse engineering western tech and manufacturing cheap generic clones of it, it's the reason why we are seeing these huge tariffs now in the us and why americans aren't allowed to buy chinese electric cars which are much cheaper and better than the soykafty overpriced crap that tesla and other western companies are putting out, which are ironically also manufactured in china but they get special exemptions from the tariffs. anyway, china has already reverse engineered western ai models to make their deepseek model and they will undoubtedly be putting this machine learning technology to good use in their future reverse-engineering efforts, and i wouldn't be surprised if india eventually starts doing the same thing as well. ai is like kicking dead whales down the beach at generating original ideas, but one thing it is quite good at is combing over lots and lots of other people's existing ideas and filling in the blanks and it will only get better at doing this as the technology evolves. we may very well be looking towards a future where intellectual property ceases to have any real meaning and ip law becomes effectively unenforceable, when any product that is released to the market can be dissected and deconstructed with the assistance of machine learning tools and refactored into a functional generic cheap clone. software patents, pharma patents, closely-guarded industry trade secrets, all of it might be going bye bye and there's not really anything anybody can do about it. personally i'm kind of excited about this, i think patents and intellectual property are bullshit that stifles competition and innovation and leads to monopolization of technology. proponents of patent law try to say they do the opposite, that nobody would ever bother researching or inventing anything if not for the ability to patent their discoveries and lock everyone else out of using or improving upon their ideas, but i think that's all bullshit. and at any rate, it's not like all tech innovation would cease, it would just be done by the public sector instead of private sector, and considering the kinds of things that the private sector devotes all their research to that's probably not a bad thing.

>>91765 >How effective would a new license be? I'm thinking clauses that forbid using AI to create "legally distinct" code. the problem with that is the original authors of most software projects are already using ai to write their code to some extent or another and there's not really any clear way to draw the line between what constitutes a human-created work and what doesn't. here's an interesting article i found about this: https://medium.com/@arturormk/how-ai-ends-intellectual-property-without-trying-the-collapse-mechanisms-b27f3bdf2e61

Please tell me, does this affect the original copylefted code negatively? Corpos and suits will have another code with the same functionality, so what? I never gave a fuarrrk about copyright or copyleft, everything is public domain in my eyes, let normalworms fight over their stupid laws as long as they want

>>91768 gpl proponents say that public domain/permissive licenses allow corporations to "steal from the commons" by using open source code and not contributing their changes back and restrictive copyleft licenses like the gpl are meant to prevent this, but now with ai corporations can "steal" whatever they want by simply saying "oh we didn't copy from the gpled code, we copied from this public domain ai-generated fork of the gpled code, which we definitely had no part in generating and good luck proving otherwise"

>>91769 Yeah i understand that much, but i mean is "stealing from the commons" affect "the commons" any negatively? We can still use and develop our copyleft projects, right? It's just retards will pay money for a gimped and glowniggered version of something that they could have got cleand and for free, right?

>>91770 >Yeah i understand that much, but i mean is "stealing from the commons" affect "the commons" any negatively? it might deter ideological purists from contributing to open source software because they don't want to end up being unpaid indirect contributors to proprietary software projects that use their code, but i say good riddance, who needs them. apple used freebsd code for macos and freebsd is still going strong, and i seriously doubt any of apple's proprietary code would be useful to anyone other than apple anyway.

How good is its Minecraft clone?

>>91762 So could we now take leaked sources and make them legally distinct enough to be officially used? Can this potentially rewrite, say, Windows XP? Would be kinda fun.

>Yeah i understand that much, but i mean is "stealing from the commons" affect "the commons" any negatively? It allows corpos not to contribute to the commons. Maybe they wouldn't have done so anyway, but then they would have had to develop their own in-house alternative from scratch (or buy it from someone else who did), reducing their margin or competitiveness. So corpos taking from the common without giving back isn't bad in itself, just like piracy isn't stealing, but it still isn't behavior that should be encouraged in society.

If the law were sensible, this would also mean that any LLM generated code or other content is public domain and therefore not copywriteable either or distributable as proprietary software. There is an extremely strong case to be made here, but the law is not really a code of rules. Rather, it is the domination of one class by another, and information wants to be free.

It's satire https://gigazine.net/gsc_news/en/20260313-malus-open-source/#gsc.tab=0

AI was planned long in advance, a major reason the internet even exists is to create ai training data. The internet began as DARPANET from the US Department of Defense, facebook is darpa's lifelog, google was funded by the nsa, etc. The feds orchestrated the open source movement to produce a supply of ai training data, the snake eats its own tail when it is no longer needed.

>>91776 >If the law were sensible, this would also mean that any LLM generated code or other content is public domain and therefore not copywriteable either or distributable as proprietary software as i already said twice now, existing copyright laws have already declared that machine-generated work is not copywriteable

>>91782 >existing copyright laws have already declared that machine-generated work is not copywriteable that makes no sense. so, i can just steal your code and say that an LLM made it?

Well there's a space between 'copyrightable' and GPL style copyleft. Public Domain code can afaik be used in 'proprietary' software no problem, and I don't think you're obliged to release your code if its public domain (afterall, its already released). The main thing is that you can't go after people using the same code and call them thieves. The more amusing argument that I'm sure stallmanfans would make is that because inevitably these AI are trained off GPL code and can reproduce code licensed GPL, its reasonable to say that GPL copyleft requirements apply and therefore everything coming out of one of these LLMs is GPL and any project using it must also be GPL.

OP's site is satire, but this is a real thing. https://antirez.com/news/160 I'm not sure how the "clean room" argument would actually hold up in court. In step 2, it is missing the crucial part of checking that the "specification" doesn't contain any copyright-infringing data. In step 5, an LLM claiming that the code is clean is no argument at all, as they are very good liars. >>91783 >that makes no sense. so, i can just steal your code and say that an LLM made it? No, because that is false. The colour of the bits matter. http://n4celbknwkn4twryohqzdko3txv7p3s7kgalvdglapd74vg3yrzin6id.onion/~jeff/files/mirrors/ansuz.sooke.bc.ca/entry/23.html (file-rel, if you can't use Tor for some reason)

>>91790 Re: antirez article Even if the specification was properly inspected in step 2, the LLM's output is still tainted by its training data.

>>91784 Laws only apply to peasants, that's the flaw with the argument.

>>91790 Wouldn't this also mean that clean-rooming some code doesn't really change anything

>>91793 What do you mean? Clean-room reimplementations work because the clone is written by people who never saw copyrighted code.

>>91792 this isn't about laws its about contracts and contracts get broken or half broken or twisted in weird ways all the time. Lots of contracts are legally unenforceable and even ones that are are often not enforced.

>>91762 Now let ai reverse engineer closed source code, rewrite it, and poof, public domain ;)

>>91762 > you can now take any code and turn it into public domain that's the holy grail of software freedom however

>>91790 In $CURRENT_YEAR I would say vibecoding in a blessing and reverse engineered BIOS via using AI is based too lol imagine when new boards are reversed jk that soykaf is boot locked in the dyes

>>91772 Luanti? it's good despite the gay as fuarrrk name. Minetest was fine.