A new twist in the "AI license laundering of chardet" story https://github.com/chardet/chardet/issues/327

[soapdog@toot.cafe]

@cwebber that whole relicensing and this slop reply are vomit inducing.

[dajb@social.coop]

@soapdog @cwebber It's just the lack of understanding of what an LLM is that's makes one's hand want to smack one's forehead. Or, preferably, his.

[rcriii@hostux.social]

@cwebber I love the sentence "If you are indeed the Mark Pilgrim..." So steeped in bad faith that you assume others are too.

[cyberia@tilde.zone]

@cwebber Well, the maintainer's point was that this is "clean room", by which they mean Claude was not given the existing codebase as input. The counter argument is that the existing codebase almost certainly forms part of Claude's training data, so the claim of it being genuinely clean room is bogus. So to make your idea work, you'd have to use the proprietary codebase as training data, rather than prompt input.

[cyberia@tilde.zone]

@cwebber and I suspect that if you made an LLM based on the specific code as training data, a court would probably rule differently to how they have ruled about LLM generated code in other cases. maybe.

[hashbangperl@hachyderm.io]

@cyberia @cwebber it would need a controlled clean-room training data and training and context, so yeah it was trained on the original GPL code and is not a clean-room implementation

[npdoty@techpolicy.social]

@cwebber I cynically fear that the likely outcome is that proprietary copyright holders with lots of lawyers and money could succeed in preventing re-licensing as open source, while copyleft advocates with few resources couldn't actually prevent re-licensing to closed.

[cstanhope@social.coop]

@cwebber I think you're going to need one hell of a kickstarter to fund that one.

[cstanhope@social.coop]

@cwebber I'm not sure that's slop, but I won't discount the possibility... But this part is funny in the dark humor sort of way:

"...explicitly instructed Claude not to base anything on LGPL/GPL-licensed code."

So, you see, no problem...

[ectopod@hachyderm.io]

@soapdog @cwebber There is a real issue with people using LLMs to try to brute force their way out of a situation. Make a response that is long enough and plausible enough, and people will roll their eyes and often just give up. I have experienced this directly at work, and it drives me crazy.

[haste@mastodon.social]

@cwebber I love the idea of weaponizing their reasoning in support of the working class.

Cynically though, I think there’s a third outcome: rules for thee, but not for me. In which Microsoft uses the full weight of their wallet to crush the common person, but is free to steal themselves, to profit off of the open source community. The rest of us are left to victimize each other with little legal recourse.

Is it logically consistent? Nope, but that’s the weird timeline we live in.

[aeva@mastodon.gamedev.place]

@cwebber these people don't know how to write on their own anymore lol

[kirtai@tech.lgbt]

@cwebber
If he can't be bothered to write it, why should we bother to read it?

[msh@coales.co]

@cwebber I think the only sticking point with this scheme is the concept of a vibe coded "clean room implementation" is problematic. Like, have you SEEN Claude's room? Is absolutely FILTHY!

[vonubelgarten@mastodon.sdf.org]

@cwebber even funnier with *closed source* proprietary Java or C# apps (and Android, perhaps?!) as these can be decompiled to a very ugly IR code that can be somewhat usable to guide a LLM!

[sprocketclown@mastodon.social]

@cwebber What constitutes laundering of copyleft codebases?

[gumnos@mastodon.bsd.cafe]

@SprocketClown

The way I read it in this context is that an existing codebase has license (whether GPL, LGPL, or proprietary or whatever), and that by "laundering" the codebase through an LLM, the output no longer retains the retains the license terms. In the US at least, the Supreme Court has ruled that LLM output is uncopyrightable.

So as @cwebber highlights, either the licensewashing works, in which case LLMs can scrub licenses off proprietary codebases giving a leg up on "reproducing" proprietary codebases into the public domain; or it doesn't work, in which case LLM-produced code becomes subject to the licensing of the original code.

[feld@friedcheese.us]

@cwebber

> Their claim that it is a "complete rewrite" is irrelevant, since they had ample exposure to the originally licensed code (i.e. this is not a "clean room" implementation). Adding a fancy code generator into the mix does not somehow grant them any additional rights.

The human didn't write the code, the LLM did. "They" which had "ample exposure to the originally licensed code" does not exist; "they" are ephemeral.

1. Start a fresh session / clean context, make it meticulously document the architecture, APIs, etc

2. keep those documents, throw away the code, start a new session with an LLM that has clean context and tell it to build off those documents.

That's clean room. If the original code was not in the LLM's context, it's not violating the license.

This is how you can do this. Proving beyond a reasonable doubt he didn't do it this way is going to require a lot of evidence nobody will have.

[feld@friedcheese.us]

@cwebber how is than an "obvious slop response"? I don't see anything odd other than the "core claim" statement but I would probably have phrased it similarly

[cwebber@social.coop]

@feld The headings, the emdashes, the framing of sentences, all classic AI "speech patterns" especially in markdown documents