You must log in or register to comment.
Also, all this reminds me of drama in the Skyrim and Minecraft modding scenes, when devs publish stuff under Apache or MIT or whatever.
Then the devs find out they don’t like what others are doing with their code. Drama ensues.
…That’s kinda the deal with permissive licenses. Or posting publicly, like here on Lemmy. People will do things you don’t like with your code or content.
Nah if you’re a massive AI company and you scrape without contribution, you’re a huge piece of shit.
It’s just stealing plain and simple like anything else.
They’re not a small user getting open source software, they’re scraping what is already done to try and make you obsolete.
There are permissive licenses, and then there are copyleft licenses. Permissive licenses go in the direction of “do whatever the fuck you want”. Copyleft licenses are more like “use it for whatever the fuck you want but if you change it give it back to everyone else with the same conditions”. The people who have projects with copyleft licenses are the ones who are (rightfully) pissed about their projects being used to train AI.
Huggingface isn’t violating copyleft licenses here, I don’t think. And research projects that use it, with citations and documentation, wouldn’t either.
Now, if some business comes along and tries to make proprietary code derived from the dataset, that’s where things get hairy. But the people doing that are responsible for the potential violation.
On the other hand, totally did not see this particular shit coming. I wonder if we’ll see a rash of “permissive, except LLMs can fuck right off” licenses.
Right, I mean I literally published my code openly, so idc lol
“Look, I wrote this neat highly advanced machine vision thing to help people with accessibility needs communicate with loved ones! ❤️. MIT licensed I guess! Let’s make the world better!”
Raytheon, Northropp, Boeing, three-letter-agencies suddenly fork it as a base for their own “projects.”
😐
There’s some stuff I also would like to work on adjacent to these things but the risk outweigh the good for as long as these psychotic people are in charge (and some things will just generally always be risky).
Institutions with monopoly on violence or companies that they deal with don’t care whether software is released as MIT or AGPL, they just take it and build on it. (Maybe that’s exactly what you mean though, I’m not sure.)
Hehe yeah I guess that’s what I meant. Just making a statement on how software has become I guess, and maybe a bit about how highly permissive licenses might have unintended consequences.
The open source spirit is one of the shining jewels of the human race, but it really sucks that I don’t think any of those “You aren’t allowed to use my code to make the world a worse place” clauses would actually hold up. =\
Well… I’d rather the dataset be public and there, with an ostensible centralized opt-out, instead of every AI startup frantically rescraping the same things their predecessors did.
True, I understand why this is better than all those companies scraping it individually (for both ability to opt out and site load), but the way they handled opt out is still quite silly.
Second that. After all all my open repositories are all either licenced under GPL or MIT, so complaining would be a kind of hypocritical.
GPL is dodgy to be included in training data for AIs that are then used to generate non-openaource code tbh. Even MIT loses the attribution it’s supposed to have once laundered through AI…
Speaking as someone that’s been into FOSS for a long time, it does piss me off how quickly copyright got thrown under the bus the moment it became inconvenient for people with money.
Yes, but what HuggingFace is doing here is the distribution of a data set. And so long the data set itself is open that doesn’t conflict with GPL. HuggingFace is not responsible about how others use that data set.
Whether LLMs themselves violate copyright for being trained on MIT or GPL code is an entirely different discussion.
If they only include repositories that come with a proper open source license technically they don’t even need to provide an opt-out option. So, good thing that at least it exists.
EDIT: oops replied to wrong comment
Genuinely surprised it’s even opt-out.
These companies train on Disney DVDs. Permission is not a factor. Training is transformative use, as much for counting letter frequency as for building a chatbot that can sort of code.
no, officer, you misunderstand! I’m not pirating this movie, I’m just training my intelligence on it! it is transformative use, see, I can now write you this summary!
Quoting one sentence from a book is fair use even if you shoplifted that book.
Get off of Github if you think this is a problem 🤷 There are alternatives like Forgejo (Codeberg), Gitlab, and Radicle (decentralised).
Those are obviously scraped as well?
Sure, but they can’t discover them all. Especially radicle is not easy to crawl, due to it’s decentralised nature. It can even be hosted on TOR and I2P
I don’t see a problem as long as they stick to AGPL when building a product out of it
Edit: Oh they also scraped my proprietary code 🧐
So they scrapped stuff that wasn’t permissive? Interesting… though not really surprising.
If someone wants to scrap the shitty ass code I have on GitHub, have at it. Talk about poisoning AI
“our project”
A number of my repos are listed.
But the weird part is that it also lists a repo I don’t recognize. The repo does actually exist on my github account, but it’s marked as “ignored”, and the description says it was automatically exported from Google Code. The code seems to be a MacOS shareware file encryption tool called “BitClamp”, published circa 2008.
No idea how it got on my account.
I’m just realizing that I’ve had my college transcripts on GitHub since I graduated years ago.
should I activate my trap card now or later?
Rip I’m in there
Out of my 33 public repositories, they only scraped the 7 oldest. Weird.
I have 3 repos in it but I don’t really care, it was all repos I had publically shared and open anyway
Wow, got about 9 repos of mine there and a few from my startup’s, only public ones are scraped so I say fair enough. Happy to have switched to running my own git infra a year ago
Thats why you write your projects now, download all weights locally and never look back.
For posterity huggingface is terrible. If you need it. Even codebergs new changes are a disaster in teh making The days of “if its on the internet” are now your own responsibilty.
Love how things are being rewritten on the fly now… Even better, lets scrape and “selfimprove”… WhatCouldPossiblyGoWrong
In theory, it’s all supposed to be permissively licenced code and the opt out is more than other models give. I saw Starcoder as one of the more ethical models. I thought the underlying principals to be fair at least.
I’m interested in this gut hostility to it regardless. Kind of shows how you can’t present LLMs in a positive angle no matter what. They shouldn’t be using anything GPL or similar.
Kind of shows how you can’t present
LLMsharvesting peoples data without consent or even warning and making it difficult to impossible for people to avoid it in a positive angle no matter what.Fixed it for you.
An LLM built from only consensually provided data would be perfectly fine, as long as it works without environmentally ruinous data centers.
In fact, that was how EVERY LLM was to begin with, until regulatory capture and corporate impunity reached the current crescendo.
But if it’s permissively licenced, couldn’t I just copy bits and pieces for my own project without asking?
Like I understand asking is always better and an opt in process for “the stack” or wtv would have been better received. Nevertheless, did they really have a legal obligation, rather than moral obligation, to ask given how this is licensed?
Taking GPL code would be a different situation right? They would have to also include it in their model creation… which theoretically they just could.
Most of those “permissive” licenses require redistributors to redistribute copies of the license texts in derivative works.
But I bet these AI models aren’t doing that. And it’s a damn neat certainty that the vibe coders who use the AI model are not attaching a license disclosure containing every permissive licenses in GitHub. Even if their vibe coded app is arguably a derivative work.
…do you know what the word “consensually” means? 🤦🏻
I do not believe a permissive license has any notion of consent. You can’t stop someone from forking your code as long as they follow the rules of the license, like crediting you.
Even in GPL, you can’t stop someone from using your code, see Gnome’s recent arguements with Mint over their usage of an old version of their Calender app.
Others in the thread have mentioned we’ll see permissive licenses with exceptions for Ai in the near future. It’s a solution because that door is currently open.
I do not believe a permissive license has any notion of consent. You can’t stop someone from forking your code as long as they follow the rules of the license, like crediting you.
Does the LLM ever credit the original author when it spits out code?
You can’t stop someone from forking your code as long as they follow the rules of the license, like crediting you
In that case,though, you’re still respecting the wishes of the person applying that permissive license consensually, though.
That’s VERY different from mass harvesting all data without permission (or credit) for profit, which would probably be against the terms of even the most permissive licenses.
Their claim was they followed the licenses and only used code they were implicitly allowed to use. I’d like to see if that was just astroturf and bullshit.
Honestly, though this is one of the reasons why people prefer copyleft and avoid permissive licenses, because yeah anyone can profit monetarily off your work otherwise.
Their claim was they followed the licenses and only used code they were implicitly allowed to use
Was likely bullshit. Just like just about everything else people behind for profit LLMs say about their business practices.
I’d like to see if that was just astroturf and bullshit.
Almost certainly
You can host copyleft as well as all rights reserved code on GitHub. It’s not like Codeberg.
Yeah, they aren’t supposed to scrape that stuff. That’s kind of where I’m wondering if they limited their collections.
Well, they included some MPL 2.0 repos of mine at least, but skipped others that use GPL3. Yet another that is a mirror of some otherwise lost firmware files for early 2000s wifi cards (and definitely isn’t free software) is also included.
So possibly they filter out GPL2/3 specifically, rather than only include known permissive licenses. Which is a pretty bad way of doing it.
Machine learning is more than just “transformative use” and is not copying. Currently that is only like 98% true, because memorization does occur in a few cases. Like 0.8-2% and that number is probably less now 2 years later than that study. Ultimately once they fix the memorization issue and “purge” these memories and can no longer reproduce licensed code (which is mostly textbook examples and boilerplate code or very popular code) this will be true transformative learning.
Then they do not require any more permission to read and learn from a book or from code than a human would. As long as you own a book or have the right to read something, you’re allowed to do whatever you want with the knowledge you gained.
