Charlie StrossIf your corporation's business model relies on lawbreaking, your corporation has no legal legitimacy.
We don't let narcotic cartels and trafficking rings list themselves on the stock exchange: why should OpenAI or Facebook be any different?
NiceMicro@cstross copyright doesn't give you total control over your creative work.
under US and Canadian law, there are "fair use" exceptions, and "being transformative" is one of those, meaning, no consent of the original author is required. And as far as I know, most copyright lawyers expect "training an AI" as transformative.
So, artists are actually demanding a change to the law in free use, while the drug cartels are breaking law that already exists.
@nicemicro Firstly, there's existing court rulings in the US that the output of AI can't be copyrighted. Secondly, I'm not American: I'm in the UK, and "fair dealing" (here) is much more restrictive than US "fair use". Thirdly, OpenAI, Facebook, et al acquired copyrighted works illegally—by hoovering up warez sites—which remains illegal regardless of "fair use" doctrine.
@cstross the first point is irrelevant, the fact that the end user can't copyright the work, has no bearing on whether the creation of the tool broke any laws.
to the second point, I'm not in the US either, and in my country of residence it is also more restrictive, but the corporations are there, so how one would litigate it internationally is interesting, but kinda ruins the drug cartel analogy (like you want to go after american weed shops because weed in the UK is illegal?)
@cstross the third point is also a good point, but let's remember, that as a user *downloading* from warez sites is also considered by most countries to be "okay", because the illegal thing is not getting duped by someone selling you contraband, the real issue is distributing the contraband. This is why (and sorry, on the internet one hears mostly US examples) the IP holders go after torrent users: on torrent you also become distributor.
@cstross so I guess if one could prove that OpenAI and Facebook did use torrent and did seed whatever they torrented, that would be interesting.
The international aspects are also interesting from a nerdy point of view, I'm curious how people litigate IP laws in the age of the internet when you can host content from anywhere to everywhere.
Because in the '80s, in the Eastern Bloc, you were allowed to infringe on patents even from the West, but not on patents from the USSR 😂
@cstross 100% off-topic, but when I was working in a pharma company back in my home country, people told me how when in 1990, the socialist economic system was switched to capitalism and the government became democratic, there were multiple processes that had to be re-engineered because they were a 100% copy of western patents, and later they had to prove that they don't use that old process anymore in court to be allowed to sell in western countries.
Ur Ya'ar@nicemicro
> And as far as I know, most copyright lawyers expect "training an AI" as transformative.
Can you back up this claim?
I was under the impression that the debate on whether AI training falls under "fair use" hasn't reached a consesus yet.
@yaarur @nicemicro Also "fair use" is US law; outside the USA, laws differ (there's no "fair use" in the UK, for example—"fair dealing" is rather more restrictive).
@cstross @yaarur yes, it is true, and I have not seen any real legal analysis W.r.t. to UK or Canada law, so I can't comment on that.
But if you published your work on a platform in America, and it was used to train an AI by a company seated in the USA, then I doubt that the British law will have much relevance in the case, as the US based AI company might argue, that by "exporting" your work to the states, it is governed by laws there.
but I'm not a lawyer, so I have no idea if this stands.
@nicemicro @yaarur Wrong. The big AI companies have been downloading pirate websites with illegally acquired ebooks, so no legal license for use of any kind. These also include UK editions taken without consent and some of these sites are hosted in places like Russia. Finally, my works *are* published legally in the USA, so I suspect the legal departments of Penguin-Random House and Macmillan might like a word with them …
@cstross @yaarur last time I checked, downloading from pirate websites in good faith is not illegal. Sharing pirated stuff on websites is. This is why copyright trolls will upload their bait to torrent sites, because they can only have a legal case against you if you start seeding back the (i.e. distributing) the stuff you downloaded.
If you (or your publisher) end up suing them, I wish you good luck, and please let us know, because I'm honestly curious how such a case would be litigated.
@yaarur @cstross Sure, it is not a consensus, the one I read in detail was the one commissioned by the Free Software Foundation to analyze the legality of Microsoft's Copiot on Github.
Maybe saying "most" is not warranted, but I have not read or heard any serious analysis by actual lawyers which said "not fair use", but this just might be based on bias.
In this write-up, section "B" has the Fair Use part.