How can I add a simple requirement “do not train Al on the source code of the program” to AGPLv3 or GPLv3 and thereby create a new license?
Don’t know is it a good place for such a question but I try :).
Why did I come up with such an stupid idea? There have been reported cases where artificial intelligence such as Github Copilot has been trained on many open source and free software projects, and in some cases it can output code snippets from GPL-licensed projects without specifying it. https://www.pixelstech.net/article/1682104779-GitHub-Copilot-may-generate-code-containing-GPL-code
I am not a lawyer, and I do not know where it is better to insert such a requirement. And how to formulate it in the best and correct form.
I understand it maybe complicated to check, to comply with this requirement and it may cause other difficulties, but I still think it can be a useful addition.
How to fit it with the fundamental freedoms of the GPL or it is unfitable?
I understand that this would make the license non-free, since it puts constraints on what the code can be used for. It’s sad that it doesn’t combine in some way. Maybe change requirements to do not train “closed source AI”(without code and training data of AI model publicly available).
And how can I name it? Is it better to name it without “GPL” If this new license cannot be considered free? NoAIFL or your variants :)?
Is it good to just add a new item?
For example like this:
Additional Clause:
You may not use the source code of this program, or any part thereof, to train any artificial intelligence model, machine learning model, or similar system without explicit written permission from the copyright holder.
or
Section [X]:
Restrictions on AI Training You may not use the source code of this program, or any part thereof, to train any artificial intelligence model, machine learning model, or similar system without explicit written permission from the copyright holder.
What you think about it? Maybe you already know licenses like this?
You can not just add terms to the GPL. That is not how it works and it is explicitly forbidden. The thing to realize is that if an AI spits out your code it must be under the same license. Restricting someones use of your code violated the 4 freedoms and makes your code non free.
I doubt GitHub will make an effort to avoid violating less common licenses anytime soon. Once you have found a license that works for you, the best thing you can do is find alternative hosting for your code.
But what will stop them from train copilot on code from other publicly available hostings. Are there any restrictions why they won’t be able to do it using something like the principles of fair use as an excuse in this case?
What will stop them is complexity and effort. Legal risk has proven to not be enough yet.
The legal system takes years to come to conclusions
I understand that this would make the license non-free
You can potentially get around that by just specifying that any AI trained on it is considered a derivative work, and thus must be released under your new license.
That said, it’s potentially moot, the argument the AI companies use for training on commercial data and art is that it’s fair use under various exemptions.
Right now there is a huge legal question surrounding copyrighted works and AI. Since the GPL is enforced via copyright it depends on the court rulings around AI.
Basically there lots of unanswered questions at the moment
That’s not up to OP to “specify;” either it already is the case (for everybody) or it isn’t, according to the legal definition of “derivative work.”
(I take the position that it is, BTW – AI code generation is massive copyright infringement in general, and a way of laundering copyleft code for proprietary uses in particular.)
I think so too & have made that point in the past.
Does anyone know of some more legally credible references that agree with us?
That’s a good idea. Now I have to think about how to formulate it better and what it will mean. :)
Don’t do it if you like the (A)GPL or free software licenses. It’ll probably void the license. At least adding clauses immediately makes it incompatible with other open source software and take away user freedom. It’s generally not recommended to do this. (Same applies to the commons clause and other additions, they usually tend to make our life as the free software community worse.)
And it would only do harm, without any benefit. The companies claim training AI is fair use. And they’ll continue doing it anyways. At the same time every hobby programmer who forks your project on Github and wants to contribute will be in breach of your license, as Github is known to feed that into Copilot…
I’d recommend taking a step back and re-think this. Why are you giving away your project in the first place? Do you really care how people use it? If yes, your interests aren’t completely aligned with the idea of Free Software, which grants the user 4 essential freedoms. Including using your project for arbitrary purposes. As the author, it’s your decision. But the (A)GPL isn’t really the correct license/vision if you don’t fully agree with the premise.
Yes I understand It’s complex question. In principle, I support the freedoms declared in the GPL. But the GPL license itself restricts the use of code in closed source proprietary programs for the sake of the freedom of all future users. And the question arises, isn’t the whole point of this nullified if you can train an “AI” model on this code, and then use the output from the “AI” of the same code in closed sourced proprietary programs? I wouldn’t mind if these “AI” were the same kind of free and open source software, but even then you can use their output to create your own closed source proprietary programs… Maybe you are right, it is not entirely clear what is better in this case.
I mean I can also read your code, take inspiration from that and use it to write some proprietary software. I think that’s at least somewhat similar to what happens with AI. AI doesn’t reproduce the code verbatim, but instead learns from it and as far as I know nobody found it repeating large chunks of one specific software. So I’d say it’s like me reading copyrighted computer science books, learning programming that way and nowadays using my skill to code Free Software (or whatever).
I know this position is disputed. But I think there is some truth to it. At the same time it’s close to the tech companies’ rationale. It’s morally wrong that they get to profit from other peoples’ labor. And they’re definitely exploiting the situation that law and licences come from a time, where AI wasn’t an issue… But I’m really split on the topic. Ultimately we’d need some consensus on how to handle this. And some laws and regulation. And we don’t have that yet.
And I think it’s also similar to other companies profiting off of FLOSS projects. Like with Redis, MongoDB(?) and all the projects that shifted from open-source to source-available due to Amazon et al just taking things and making profit by selling it as a cloud service without ever contributing back. Is just a sad situation. And ultimately it harms me and everyone. Because I’m subject to the same license. And now I can’t use, modify and share some software anymore. These non-commercial clauses are difficult, too. Even if I just run a small Fediverse instance and collect donations, that could be construed as commercial. Or trying to make a living off of Free Software. And I think all of this drama is an even bigger problem than AI being trained on other people’s code. And it all cuts down on freedom. I mean for a legitimate reason… And I get it… Still the freedom gets lost.
Thank you very much for your reply. I support your opinion in a way that I am already inclined that a complete prohibition on the training of “AI” models on the source code of software is not a very good solution and is difficult to limit according to current laws. I hope somtimes someone smart will come up with some approaches to such problems.
Indeed. I hope so. And we desperately need some clear regulations. Even the big AI companies struggle with the lack of clear rules. I can see how we need to go through quite some legal battles to settle some questions arising with the new technology. And that’s currently taking place. But it extends past that. Currently, companies are retreating from the European market. Due to a completely unmanageable situation. I’ve seen local language models (starting with Llama 3.2) being banned / not licensed within the EU. And that’s going to lead to all kinds of complications. Just because the EU can’t get some proper regulations out, and do it in time. That’ll leave technological progress behind in the EU, mess with companies. In effect also take away my freedom to run language model on my own hardware…
I hope they get that straight. And there is some demand… So maybe it’s happening sooner than later. But these are very difficult questions to answer. About AI safety, copyright, effect and impact on society and freedom… And I think a lot of these questions are difficult to tackle with licensing anyways. We definitely need laws governing if AI training is fair use. Or if generating a voice that sounds 70% like David Attenborough is alright to do.
Do you really care how people use it? If yes, your interests aren’t completely aligned with the idea of Free Software
I wonder how those proposals for a new “Ethical Free Software” movement and licenses are doing, since one of the simplest rational expectations when you release a software is that the user won’t use it to harm you.
My position is that AI code generation already violates the GPL if it was trained on GPL code.
(Unless the AI output is itself GPL-licensed, in which case it merely violates the license of anything that isn’t compatibly-licensed.)
The question is what qualifies as a derivative work
