Hacker News Reader

Comment by lelanthran | Hacker News Reader

You dont know what GPL is?

It's not an unconditional gift, it's got strings attached.

AI training on GPL works is basically IP laundering, you're taking the product without paying the asking prices.

IP as a concept has always been equal parts dystopian and farcical, and efforts to enforce it have become increasingly strained over time. Property requires scarcity. Ideas aren’t scarce. My consumption of an idea is affected by your consumption of an idea.

AI has simply increased the intensity of this friction between IP and reality to a degree that it can’t be ignored or patched over any longer.

I do know what it is, I've even read the licence in full!

What specific paragraph in the GPL prohibits training of AI on it? I guess it might be a matter of interpretation, but by my reading, it is allowed.

Ps. In the future, try to refrain from using demeaning rethorical questions like the one this comment starts with, it only serves to foster toxicity. Please and thank you Ds.

> What specific paragraph in the GPL prohibits training of AI on it? I guess it might be a matter of interpretation, but by my reading, it is allowed.

It's not a matter of interpretation - any derivative product is also GPL, and if you don't want the derivative product to be GPL, then don't use the original product.

Is reading source code using it? Can you restrict people from doing that? What actually makes a derivative work.

Can I put up a sign with a fact on it, can people who see the sign not use the fact unless they agree with my terms and conditions? That certainly would be the case if we went wiTh some sense of derived.

The law needs specifics for a reason, if it were down to what each individual felt it means in the moment it would be useless.

The most recent legal findings have said that training on legally acquired data does not violate copyright.

Facts aren't subject to copyright.

As for what constitutes a derivative work, this is a matter of law. In the US,

A "derivative work" is a work based upon one or more preexisting works, such as a translation, musical arrangement, dramatization, fictionalization, motion picture version, sound recording, art reproduction, abridgment, condensation, or any other form in which a work may be recast, transformed, or adapted. A work consisting of editorial revisions, annotations, elaborations, or other modifications which, as a whole, represent an original work of authorship, is a "derivative work".

(17 U.S.C. § 101)

Good point about facts. It applies similarly to properties that provide functionality. One tends to lose track of the fact that much of software shouldn't be copyrightable in the first place, it's just a pretence that has evolved due to how much people like money.

It's a stretch to say that training a model falls under that definition of derivative work. It's be like saying that building a house after reading a book on how to build a house makes the house a derivative work. I can just imagine cookbooks introducing limited licences on who you can feed with their recipes.

So the pixel editor I made using AI that was trained on, among other things, the Linux kernel, is to be considered derivative of the Linux kernel?

And that's not an interpretation?

Reply on news.ycombinator.com