FaceDeer

AI models don’t actually contain the text they were trained on, except in very rare circumstances when they’ve been overfit on a particular text (this is considered an error in training and much work has been put into coming up with ways to prevent it. It usually happens when a great many identical copies of the same data appears in the training set). An AI model is far too small for it, there’s no way that data can be compressed that much.

DMCA is about copyright (that’s what the “C” is). The name of a show isn’t copyrighted, it’s trademarked. Different type of IP altogether.

“Takedown notice” has legal meaning, it’s not some random cease-and-desist letter that you can draft for anything you want and that has no legal weight other than that it might be scary.

Seemed pretty fair and fact-based to me. What bias are you seeing?

Yes, that would also be statistical correlations to an AI model. The specific kind of information they’re being trained on doesn’t affect the underlying mechanism of model training.

Especially because seeing the same information in different contexts helps mapping the links between the different contexts and helps dispel incorrect assumptions.

Yes, but this is exactly the point of deduplication - you don’t want identical inputs, you want variety. If you want the AI to understand the concept of cats you don’t keep showing it the same picture of a cat over and over, all that tells it is that you want exactly that picture. You show it a whole bunch of different pictures whose only commonality is that there’s a cat in it, and then the AI can figure out what “cat” means.

They need to fundamentally change big parts of how learning happens and how the algorithm learns to fix this conflict.

Why do you think this?

There actually isn’t a downside to de-duplicating data sets, overfitting is simply a flaw. Generative models aren’t supposed to “memorize” stuff - if you really want a copy of an existing picture there are far easier and more reliable ways to accomplish that than giant GPU server farms. These models don’t derive any benefit from drilling on the same subset of data over and over. It makes them less creative.

I want to normalize the notion that copyright isn’t an all-powerful fundamental law of physics like so many people seem to assume these days, and if I can get big companies like Meta to throw their resources behind me in that argument then all the better.

Remember when piracy communities thought that the media companies were wrong to sue switch manufacturers because of that?

It baffles me that there’s such an anti-AI sentiment going around that it would cause even folks here to go “you know, maybe those litigious copyright cartels had the right idea after all.”

We should be cheering that we’ve got Meta on the side of fair use for once.

look up sample recover attacks.

Look up “overfitting.” It’s a flaw in generative AI training that modern AI trainers have done a great deal to resolve, and even in the cases of overfitting it’s not all of the training data that gets “memorized.” Only the stuff that got hammered into the AI thousands of times in error.

Training an AI does not involve copying anything so why would you think that fair use is even a factor here? It’s outside of copyright altogether. You can’t copyright concepts.

Downloading pirated books to your computer does involve copyright violation, sure, but it’s a violation by the uploader. And look at what community we’re in, are we going to get all high and mighty about that?

What did I say that implied that? I’m pointing out a contradiction in kilgore’s comment, I’m not adding anything of my own here.

Their distribution of books is completely legal.

Corporations just have more money to warp the laws in their favour.

You just contradicted yourself in two sentences.

But I think the law is pretty clear, and a precedent calling their use case fair use would be mind blowing. You need new, much more common sense IP legislation that redefines consumer rights in a digital world.

Indeed. I’m a big supporter of IA’s mission, and I’m a big supporter of piracy (copyright has gone insane over the years), but this outcome was obvious from the moment IA did this and it was a mistake for them to fight this fight. They should focus on preservation. Let the EFF handle the lawsuits, and let Library Genesis handle the illegal distribution of books. Everyone focus on what they’re best at.

They’re appealing the decision so there’s still opportunity for IA to throw good money after bad on this.

The time limit is a century or so, so that’s something our descendants can figure out.

As I said, the “traditional” CDLs were also in a legal grey area. But once the publishers are suing IA for going full Library Genesis anyway, why not also include those?

I went back to one of the older articles I could find on this subject, from before the lawsuit was filed. Some particularly-relevant excerpts:

Until this week, the Open Library only allowed people to “check out” as many copies as the library owned. If you wanted to read a book but all copies were already checked out by other patrons, you had to join a waiting list for that book—just like you would at a physical library.

Of course, such restrictions are artificial when you’re distributing digital files. Earlier this week, with libraries closing around the world, the Internet Archive announced a major change: it is temporarily getting rid of these waiting lists.

…

James Grimmelmann, a legal scholar at Cornell University, told Ars that the legal status of this kind of lending is far from clear—even if a library limits its lending to the number of books it has in stock. He wasn’t able to name any legal cases involving people “lending” digital copies of books the way the Internet Archive was doing.

…

The legal basis for the Open Library’s lending program may be even shakier now that the Internet Archive has removed limits on the number of books people can borrow. The benefits of this expanded lending during a pandemic are obvious. But it’s not clear if that makes a difference under copyright law. “There is no specific pandemic exception” in copyright law, Grimmelmann told Ars.

Ironically the FAQ that Internet Archive put online has been taken down, but I found it in their Wayback Machine. It says:

The library will have suspended waitlists through June 30, 2020, or the end of the US national emergency, whichever is later. After that, waitlists will be dramatically reduced to their normal capacity, which is based on the number of physical copies in Open Libraries.

So it seems pretty clear to me that by “suspending waitlists” it means that they’re going to “lend” more copies simultaneously than they actually have.

The Internet Archive had been poking a bear with a stick for years and the bear had been grumbling but not otherwise responding. So they decided to try giving it a whack across the nose with the stick instead. Normally I’d just sigh and shake my head at their stupidity, but they’re carrying a precious cargo on their back while they’re needlessly provoking that bear, and now they’re screaming “oh no my precious cargo! Help me!” While the bear has a firm grip on their leg. That makes me extra frustrated and angry at them for doing this.

I’m not siding with the bear here, I should be very clear. The publishers are awful, the whole concept of copyright has become corrupt and broken, and so on and so forth. But the Internet Archive isn’t supposed to be fighting this fight. They were supposed to be protecting that precious cargo, and provoking the bear is the opposite of doing that.

The emergency library followed the same legal framework that ebook lending follows at local libraries.

No, it did not. From the Wikipedia article:

On March 24, 2020, as a result of shutdowns caused by the COVID-19 pandemic, the Internet Archive opened the National Emergency Library, removing the waitlists used in Open Library and expanding access to these books for all readers.

Emphasis added. They took the limits off.

What the libraries do is already in a legal grey area, the publishers just don’t go after it because it’s more trouble than it’s worth and would bring bad press. Like how most rightsholders ignore fanfiction. But the IA went way beyond that and smacked them in the face.

Don’t blame IA for fulfilling their mission to make knowledge free.

Their mission is archiving the Internet. a mission that they are putting at risk with this stunt.

I expect them to not provoke the $200 billion lawsuit in the first place. They should never have done the “Emergency Library”, it was an obvious boneheaded decision.

Then, once they had done it and the inevitable lawsuit came down on them, they should have tried to settle the lawsuit. Not fight to the bitter end, not double down. They’re only making it worse for themselves. It’s not simply losing the lawsuit that could destroy them, it’s refusing to negotiate.

Oh, for crying out loud, Internet Archive. This is not the fight you should be fighting.

The Internet Archive is the steward of an incredibly valuable repository of archived information. Much of what it’s got squirrelled away is likely unique, irreplaceable historical records of things that have otherwise been lost. And they’re risking all of that in this quixotic battle to share books that are widely available anyway and not at all at risk.

“Lending” out those books in the way that they did was blatant copyright violation spitting directly into the eye of publishers known to be litigious and vindictive. All to fight for a point that’s not part of their mandate, archiving the Internet. They’re going to lose and it’s going to hurt them badly.

Each copy can only be loaned to one person at a time, to mimic the lending attributes of physical books.

Internet Archive believes that its approach falls under fair use but publishers Hachette, HarperCollins, John Wiley, and Penguin Random House disagree. They filed a lawsuit in 2020 equating IA’s controlled digital lending operation to copyright infringement.

That is not what the lawsuit was about, Internet Archive. If you’re going to fight this fight then be honest about what exactly you’re fighting for. The lawsuit in 2020 was not about one-person-at-a-time lending, it was about your “COVID Emergency Library” where you removed all restrictions and let people download books freely.

I strongly believe that copyright has gone berserk of the decades and grown like an uncontrolled weed, harming the intellectual commons for the sake of megacorporations’ profits. I’m a subscriber on this piracy community, after all. I believe in the position that Internet Archive is fighting for here, despite all the downvotes I’m surely about to be hammered with. But they shouldn’t be the ones fighting it. Let someone else take this one on. Sci-Hub or Library Genesis, maybe.

I can’t recall the last time I pirated anything executable (games and other software). There are legitimate free options for everything I’ve wanted, and executable code is just too risky.

You’d think a literal Princess wouldn’t be so keen on revolutionary communism. Doesn’t typically go so well for the royalty.