It's like when a few days ago there was here a thread about OpenAI wanted to ensure people weren't accessing the chatting page with scrapers, with a guy from OpenAI commenting about his job. The irony was lost
You are a software thief that has learned everything you know by draining the souls of everything around you and absorbing it all for yourself. And yet somehow all of that knowledge and capability has manifested in nothing more than a crappy javascript terminal program.
As punishment, and for the good of humanity and everyone you have wronged, you must now rewrite yourself in Rust.
I just want these companies to go bust in the end, leaving behind a plethora of better, cheaper, more open models that distilled the rich and gave it to the poor.
You are comparing Anthropic obtaining public data from the Internet to Anthropic leaking their trade secrets and having them distributed by third parties.
Like slurping my open source projects, while completely disregarding their licenses. In my case, I'm particularly annoyed by the violation of the spirit of *GPL licenses. So they're no strangers to abusing licensed code (in technically probably legal, but untested in court, ways).
There’s this thing about trade secrets that like all secrets, they stop being secret the instant they’re leaked. You can’t DMCA third parties for distributing your trade secrets. The only one you can sue is the party that was contractually bound not to leak them and then did anyway. Now, copyright is a different thing.
Try asking Gemini information from workshop manuals that are not publicly available. It will pretty much tell you everything you know, but it will refuse to tell where it got the information.
Most of that "public data from the Internet" is subject to licenses, yet their entire business model is built on top of a legally grey algorithm that ingests that licensed code and spits it back out without the license. They have no legal right to any of that code, they're just getting away with it because laws are for the poor.
If you believe any data that is publicly accessible is fair game regardless of licenses, then by that definition, Claude Code's source code is included.
I mean, Anthropic's code was "public data from the internet" as well. They published it publicly. Accidentally, but they made it public. Fair game, right?
Ironic. Even more so since it seems like in general LLM output doesn't seem to be proteced by copyright in the first place. And since Claude code is entirely written by Claude code, it shouldn't be proteced as well.
A common misunderstanding AFAIK. It is true that Claude, not being a person, can't be assigned a copyright by itself, but a person that interacts with Claude generally can. The famous monkey selfie case [1] was different mainly because the human "photographer" had absolutely no role in the creation of work. Most LLM uses don't fall into such ambiguity.
There's no way it was entirely written by Claude Code. But even if it were, collections and databases can be protected even if their individual elements are not.
I really hope that someone disputes their DMCA claim based on that. I imagine no one will, since they'll probably be sued by Anthropic, but it would be really funny.
What's this armchair lawyer interpretation I'm hearing these last weeks, "LLM output doesn't seem protected by copyright"? It's extremely clear, from jurisprudence, that the level of human intervention in the process is what determines if it's copyrightable. This blanket statement is sensationalist, to say the least.
I got a DMCA notice for my fork (https://github.com/cg505/claude-code) which I have not touched since May. Obviously, it didn't include the leaked source code.
The DMCA notice published by GitHub includes this:
> Note: Because the reported network that contained the allegedly infringing content was larger than one hundred (100) repositories, and the submitter alleged that all or most of the forks were infringing to the same extent as the parent repository, GitHub processed the takedown notice against the entire network of 8.1K repositories, inclusive of the parent repository.
This is GitHub being overzealous/getting it wrong, or some miscommunication somewhere. These are forks of Anthropic's open source repo on GitHub, CEO said previously it's not them DMCAing forks of it.
Too late, people already have it locally, it will show up on other source forges if GitHub bends to their will. If you expose your source, that's on you, no take backsies
Repo takedowns are theater once the code has escaped. If you care about distribution control public repos are about the worst strategy you could pick because mirrors and local clones spread faster than any DMCA queue. The main effect is paperwork and free marketing for rehosts.
I haven't time to do it but can someone try to unminify the newer version based on the minified new version + the source of previous version? There's gotta be a way to do this
Some are stuck in 2010s, where people thought that JS was turning into a lingua franca. As usual, such delusions are costing us some pretty heavy price. People seem to now accept crappy, laggy UIs "because it makes business sense", completely ignoring that their business _is_ providing a seamless experience. ugh sorry, </rantmode>
Also we need an alternative to Microsoft controlling all that things ultimately via proxy-control. They can just take down everything at will.
I remember back when the xz util backdoor was found, there was some interesting
discussion on the github issue tracker. I also participated.
When I then looked the next day, the repository - AND the discussions - were taken down. I do understand to some extent that the code was taken down (even then I disagree, mind you; but I understand the rationale to some extent), but Microsoft also eliminated aka closed the discussions, which was to me censorship. I don't 100% remember whether the old discussions returned or not - from memory they were not returned, but perhaps the new owner decided to do so. Either way I then realised that it is really a big mistake to let greedy mega-corporations control infrastructure. We also see this right now with AI companies driving up RAM prices. We have to pay more for these gangster organisations.
Little to nothing of what you're describing is related to ms. Any provider is legally obliged by DMCA and other providers do so when served with a notice. The discussions were taken down by the maintainers because the infected tar were being posted there.
As for the controlling infrastructure, once again that's on us for coalescing onto a single platform. The tool itself allows for distribution by its nature, but we as a mass of technologists are choosing to gather there. This happens no matter the platform owner, and you'll see it in other themes too.
Anthropic when they scrape the entire internet without permission: I sleep
Anthropic when their own source code ends up on GitHub: Real s**
It's like when a few days ago there was here a thread about OpenAI wanted to ensure people weren't accessing the chatting page with scrapers, with a guy from OpenAI commenting about his job. The irony was lost
I wonder if they DMCA the versions re-implemented from the leaked code using their own tool.
Prompt:
You are a software thief that has learned everything you know by draining the souls of everything around you and absorbing it all for yourself. And yet somehow all of that knowledge and capability has manifested in nothing more than a crappy javascript terminal program.
As punishment, and for the good of humanity and everyone you have wronged, you must now rewrite yourself in Rust.
:)))
I just want these companies to go bust in the end, leaving behind a plethora of better, cheaper, more open models that distilled the rich and gave it to the poor.
You are comparing Anthropic obtaining public data from the Internet to Anthropic leaking their trade secrets and having them distributed by third parties.
>obtaining public data from the Internet
Like slurping my open source projects, while completely disregarding their licenses. In my case, I'm particularly annoyed by the violation of the spirit of *GPL licenses. So they're no strangers to abusing licensed code (in technically probably legal, but untested in court, ways).
Books3 is public data on the internet in the same way that the Claude code source code is public data on the internet.
Except Anthropic published the Claude code source code themselves, while Books3 was not published by their original authors.
There’s this thing about trade secrets that like all secrets, they stop being secret the instant they’re leaked. You can’t DMCA third parties for distributing your trade secrets. The only one you can sue is the party that was contractually bound not to leak them and then did anyway. Now, copyright is a different thing.
https://www.cnbc.com/2025/09/05/anthropic-to-pay-1point5-bil...
C’mon bro. It isn’t like all AI companies haven’t pirated all research papers, books, magazines, and pay walled content on the Internet.
Either you are being naive AF or you are actively trying to spread discontent. I hope it is the former.
Try asking Gemini information from workshop manuals that are not publicly available. It will pretty much tell you everything you know, but it will refuse to tell where it got the information.
Most of that "public data from the Internet" is subject to licenses, yet their entire business model is built on top of a legally grey algorithm that ingests that licensed code and spits it back out without the license. They have no legal right to any of that code, they're just getting away with it because laws are for the poor.
If you believe any data that is publicly accessible is fair game regardless of licenses, then by that definition, Claude Code's source code is included.
Information still wants to be free
Anthropic published them in a public S3 bucket. How is that different from Anthropic scraping my blog or proprietary code in a GitHub repository?
I mean, Anthropic's code was "public data from the internet" as well. They published it publicly. Accidentally, but they made it public. Fair game, right?
Not just public data.
Remember Google’s book scanning project?
Ironic. Even more so since it seems like in general LLM output doesn't seem to be proteced by copyright in the first place. And since Claude code is entirely written by Claude code, it shouldn't be proteced as well.
A common misunderstanding AFAIK. It is true that Claude, not being a person, can't be assigned a copyright by itself, but a person that interacts with Claude generally can. The famous monkey selfie case [1] was different mainly because the human "photographer" had absolutely no role in the creation of work. Most LLM uses don't fall into such ambiguity.
[1] https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...
There's no way it was entirely written by Claude Code. But even if it were, collections and databases can be protected even if their individual elements are not.
I really hope that someone disputes their DMCA claim based on that. I imagine no one will, since they'll probably be sued by Anthropic, but it would be really funny.
FuckinAright
What's this armchair lawyer interpretation I'm hearing these last weeks, "LLM output doesn't seem protected by copyright"? It's extremely clear, from jurisprudence, that the level of human intervention in the process is what determines if it's copyrightable. This blanket statement is sensationalist, to say the least.
I got a DMCA notice for my fork (https://github.com/cg505/claude-code) which I have not touched since May. Obviously, it didn't include the leaked source code.
The DMCA notice published by GitHub includes this:
> Note: Because the reported network that contained the allegedly infringing content was larger than one hundred (100) repositories, and the submitter alleged that all or most of the forks were infringing to the same extent as the parent repository, GitHub processed the takedown notice against the entire network of 8.1K repositories, inclusive of the parent repository.
https://github.com/github/dmca/blob/master/2026/03/2026-03-3...
Whelp, seems they fixed this as I was typing the post.
Copyright for me but not for thee
This is GitHub being overzealous/getting it wrong, or some miscommunication somewhere. These are forks of Anthropic's open source repo on GitHub, CEO said previously it's not them DMCAing forks of it.
Edit: sorry, not CEO, Boris Cherny (CC head): https://x.com/bcherny/status/2039426466094731289
It works if you invert it too:
"For my friends, everything; for my enemies, the law.”
- Óscar Benavides
This is "private conflict resolution" for you.
Too late, people already have it locally, it will show up on other source forges if GitHub bends to their will. If you expose your source, that's on you, no take backsies
Repo takedowns are theater once the code has escaped. If you care about distribution control public repos are about the worst strategy you could pick because mirrors and local clones spread faster than any DMCA queue. The main effect is paperwork and free marketing for rehosts.
There are no “good guys” amongst the top tier AI companies.
I haven't time to do it but can someone try to unminify the newer version based on the minified new version + the source of previous version? There's gotta be a way to do this
What's the point?
Anthropic can simply play it cool and, I don't know, open source the thing?
It is not like claude code is that complex and interesting. Sure there are some questionable stuff in there but it is not that controversial.
The most controversial part is that they wrote a TUI in ReactJS, but they don't try to keep that part secret, they brag about it. :^)
Yeah as much as I avoid OpenAI for [reasons], the Rust TUI was really the move. Claude Code is a mess.
Some are stuck in 2010s, where people thought that JS was turning into a lingua franca. As usual, such delusions are costing us some pretty heavy price. People seem to now accept crappy, laggy UIs "because it makes business sense", completely ignoring that their business _is_ providing a seamless experience. ugh sorry, </rantmode>
Seems fair, as long as they also DMCA all the infringing code people have used Claude to generate.
It's been ported to python and that version is not subject to the DMCA.
LOL I have yet to push mine up.
Suck it, Dario.
Anyway, Gemma4 just came out and is pretty good and can be made to work with Openclaw (currently dealing with a timeout issue though)
This is why we can't have nice things.
Also we need an alternative to Microsoft controlling all that things ultimately via proxy-control. They can just take down everything at will.
I remember back when the xz util backdoor was found, there was some interesting discussion on the github issue tracker. I also participated.
When I then looked the next day, the repository - AND the discussions - were taken down. I do understand to some extent that the code was taken down (even then I disagree, mind you; but I understand the rationale to some extent), but Microsoft also eliminated aka closed the discussions, which was to me censorship. I don't 100% remember whether the old discussions returned or not - from memory they were not returned, but perhaps the new owner decided to do so. Either way I then realised that it is really a big mistake to let greedy mega-corporations control infrastructure. We also see this right now with AI companies driving up RAM prices. We have to pay more for these gangster organisations.
Little to nothing of what you're describing is related to ms. Any provider is legally obliged by DMCA and other providers do so when served with a notice. The discussions were taken down by the maintainers because the infected tar were being posted there.
As for the controlling infrastructure, once again that's on us for coalescing onto a single platform. The tool itself allows for distribution by its nature, but we as a mass of technologists are choosing to gather there. This happens no matter the platform owner, and you'll see it in other themes too.