> Exception:
>
> * Migros Exception: Migros-Genossenschafts-Bund and its Affiliates are exempt from all commercial license requirements regardless of revenue, AI usage, or any other trigger.
Author works for that company. License is copied from https://github.com/xdgrulez/kafi which I presume was made for said company and therefore must contain the exception.
It's still kinda Open Source, but with deterrent clauses to make bigger organizations pay for what we OS coders create.
Imagine a tool like pandas uses this license. Or numpy. Then those who work on the code actually get rewarded for the work they do. And the big companies do not get it all for free. Including those who'd love to use that stuff for training their AI coding models ;-)
> * Migros Exception: Migros-Genossenschafts-Bund and its Affiliates are exempt from all commercial license requirements regardless of revenue, AI usage, or any other trigger.
The text of the license itself reads as AI slop, which is not the most encouraging. One of the first items in the license is an exception for the author's employer(?)
> "Migros Exception: Migros-Genossenschafts-Bund and its Affiliates are exempt from all commercial license requirements regardless of revenue, AI usage, or any other trigger."
The idea may have some merit, but reading the license I find it does not match the stated goals. To my mind, this kind of license should be easy to swap in the place of a permissive license like MIT or Apache, with the goal of protecting against big tech abuses. Legally protecting against slopforks seems almost impossible, but maybe still worth trying to write in.
This is noble of you but I don't really see how it can avoid the issue of agent forking and rewriting the entire codebase which is what is being done with GPLv3 right now.
The simplest sokution is to just copy the gpl/agpl and modify:
“The Program” refers to any copyrightable work licensed under this License. Each licensee is addressed as “you”. “Licensees” and “recipients” may be individuals or organizations.
To
“The Program” refers to any copyrightable work licensed under this License. Each licensee is addressed as “you”. “Licensees” and “recipients” may only be natural born humans.
If anyone else wants a licence they can buy one like every other piece of software on the market.
You're asking the right questions. The going theory as far as I can see is that training models is fair use (although it may not be fully resolved in the courts), in which case this whole exercise would seem to be pointless. If it were that easy, I have to think the FSF etc. would have been all over this years ago.
I love it. Ofc it‘s unclear if and how this can be enforced, but we MUST find a proper solution to this problem.
It also extends to images and videos. ML should not become an escape hatch for copyright evasion. We invented copyright for a reason - without it, our capitalistic system is flawed and unfair. Not in a sense that it‘s not „nice“: each system has its stress limits, and if we bias the odds continuously in favor of few, we risk instability.
Of course it can be enforced. If you're not compliant with the terms of a license you don't have a license and are a pirate. If the license is nonsensical then no one can run the software.
https://www.linkedin.com/in/ralph-m-debusmann-63204675/
Author works for that company. License is copied from https://github.com/xdgrulez/kafi which I presume was made for said company and therefore must contain the exception.
Hahah! Like Bitcoin: equal distribution of wealth - except for us that joined early, we get most of course.
No mention in the README either. Sneaky
Hey,
I've done some more work on a license that I'd like to put forward as a response to Open Source just making the richest even richer.
My response is the "Human Source License" (https://github.com/xdgrulez/human-source-license).
It's still kinda Open Source, but with deterrent clauses to make bigger organizations pay for what we OS coders create.
Imagine a tool like pandas uses this license. Or numpy. Then those who work on the code actually get rewarded for the work they do. And the big companies do not get it all for free. Including those who'd love to use that stuff for training their AI coding models ;-)
Please have a look :)
Cheers, Ralph
> * Migros Exception: Migros-Genossenschafts-Bund and its Affiliates are exempt from all commercial license requirements regardless of revenue, AI usage, or any other trigger.
And I'm just supposed to use this as-is?
The text of the license itself reads as AI slop, which is not the most encouraging. One of the first items in the license is an exception for the author's employer(?)
> "Migros Exception: Migros-Genossenschafts-Bund and its Affiliates are exempt from all commercial license requirements regardless of revenue, AI usage, or any other trigger."
The idea may have some merit, but reading the license I find it does not match the stated goals. To my mind, this kind of license should be easy to swap in the place of a permissive license like MIT or Apache, with the goal of protecting against big tech abuses. Legally protecting against slopforks seems almost impossible, but maybe still worth trying to write in.
This is noble of you but I don't really see how it can avoid the issue of agent forking and rewriting the entire codebase which is what is being done with GPLv3 right now.
Too compelex.
The simplest sokution is to just copy the gpl/agpl and modify:
To If anyone else wants a licence they can buy one like every other piece of software on the market.Can a license prevent someone from training on the source if it's considered fair use to do so?
No.
And training is currently considered fair use in the US (some court cases pending).
I am not a lawyer, tho.
You're asking the right questions. The going theory as far as I can see is that training models is fair use (although it may not be fully resolved in the courts), in which case this whole exercise would seem to be pointless. If it were that easy, I have to think the FSF etc. would have been all over this years ago.
[dead]
I love it. Ofc it‘s unclear if and how this can be enforced, but we MUST find a proper solution to this problem. It also extends to images and videos. ML should not become an escape hatch for copyright evasion. We invented copyright for a reason - without it, our capitalistic system is flawed and unfair. Not in a sense that it‘s not „nice“: each system has its stress limits, and if we bias the odds continuously in favor of few, we risk instability.
Of course it can be enforced. If you're not compliant with the terms of a license you don't have a license and are a pirate. If the license is nonsensical then no one can run the software.