3 ms·
From the article: > parts of the training data can be memorized in whole or in part by current generative models Humans do the same thing... >can therefore b
by ranger_danger 2y ago
From the article:
> parts of the training data can be memorized in whole or in part by current generative models
Humans do the same thing...
>can therefore be generated again with suitable prompts by end users and thus reproduced
Personal copies (even verbatim) are not disallowed in most countries AFAIK, the concern is regarding distribution which is a separate issue.
I would have thought it would be fair use: https://www.youtube.com/watch?v=X9RYuvPCQUA&t=41m44s https://www.youtube.com/watch?v=X9RYuvPCQUA&t=41m44s
Since the derived work is merely inspired by the original, just like a human being uses inspiration to create new things, whether consciously or not (hence the video "everything is a remix"). The only debate IMO is just how much is being "copied" exactly, and how obvious it is.
- diogocp 2y agoFair use is a concept from US law. The article is about German and European law, which differ significantly from US law in that regard.
- ranger_danger 2y agoYes I think fair use should exist in the EU as well.
- falcolas 2y agoWe don't even know in the US whether fair use applies. There are court cases out there which are trying to test those bounds. Personally, I don't think it is. Significant portions of copyrighted works can be re-created, and these models are being leveraged to make money. That makes them unlikely to be fair use. But, I'm not a lawyer nor judge.
- ranger_danger 2y agoI don't think there can be a blanket definition that applies to everything. Just like "inspiration" humans use to create all kinds of art already (even subconsciously), it's a case-by-case analysis, and one judge's subjective opinion, over how significant is too significant, for each creation that is distributed.
- falcolas 2y agoSure. But how many failed cases does it take for OpenAI or Diffusion close their doors? A failed copyright defense can result in a halt on distribution (shut it down while damages are worked out) in addition to fines and lawyer fees.
- ranger_danger 2y agoMy understanding is that those companies are not distributing their own generated content, and that user-submitted ones (if such things are even posted on their own website? I haven't seen much) would fall under Section 230 protections in the US.