Kate Knibbs reports in Wired magazine:
Against the company’s wishes, a court unredacted information alleging that Meta used Library Genesis (LibGen), a notorious so-called shadow library of pirated books that originated in Russia, to help train its generative AI language models. […] In his order, Chhabria referenced an internal quote from a Meta employee, included in the documents, in which they speculated, “If there is media coverage suggesting we have used a dataset we know to be pirated, such as LibGen, this may undermine our negotiating position with regulators on these issues.” […] These newly unredacted documents reveal exchanges between Meta employees unearthed in the discovery process, like a Meta engineer telling a colleague that they hesitated to access LibGen data because “torrenting from a [Meta-owned] corporate laptop doesn’t feel right 😃”. They also allege that internal discussions about using LibGen data were escalated to Meta CEO Mark Zuckerberg (referred to as “MZ” in the memo handed over during discovery) and that Meta’s AI team was “approved to use” the pirated material.
When I said “libgen is great because information should be free!” this isn’t what I meant… jeez
Did they seed at least?
it’s facebook, they probably issued a takedown request for all their logged peers
The pivot-to-ai writeup is out, they did seed! I assume it’s documented then.
Multinational corporations can act ethically after all.
It’s clear that they didn’t stop uploads of the torrents. It hasn’t been established in the documents we’ve seen so far that they actually had downloaders in turn. But they did clearly make the works available for upload.
Multinational corporations can act ethically after all.
I wouldn’t go that far
They can, they just choose deliberately not to most of the time.
In total honesty though, Meta had actually done some good things for Open Source. Sure, this is probably it of their own interest and neither outweighs nor make up for all the bad. But they can, and sometimes do.
no way, that’s illegal!
Nice! Now simply fine them to pay significant royalty to every author in there, say, a millicent per word of everything they’ve generated before they get caught.
We should just start a meme movement that makes up an imaginary yet believable fact, like the lemmings jumping off a cliff thing, wait for the ais to repeat it and lobby for royalties. Do one for each of the major ai platforms - openai, reddit, meta, apple, google etc. we would eventually find out which public forums are training which bots.
You don’t need that, all of them use everything
Doesn’t even have to be believable, LLMs Don not care.
And yet these are the things the investment bankers expect to take us to the next level lol
So as libgen is blocked here in .nl by various providers (mine calls it thepiratebay for some reason), i look forward to all their llm being blocked.
I used to think they’d just train on every Facebook account that was ‘deleted’, i.e. removed from the public eye. This feels much worse.
