Technology

They also love to publish stuff exclusively through Amazon so libraries won't get it.

"caught" also implies they've been hiding it... It's been out in the open how they scan books to train AI for a while now.

Whistle-blown?

I wish there were any path for accountability anymore.

No it only applies to books because books have an entirely different legal system than everything else.

SCOTUS case: White-Smith Music Publishing Co. v. Apollo Co. (1908) created a new issue for the Government about copying people's printed works.

Copyright Act of 1909 is passed. New framework is established, books get an entirely different legal system from everything else. Publishers lobbied to prevent it's passage and fought it especially hard in the Great Depression because libraries were becoming very popular during that time.

But printed books have a whole different thing that applies to them that doesn't apply to anything else. So you can't use "physical medium" because books are a special case and have their own special US law.

For books, transformative, has to deal with the intent.

• Book - I want to read them.
• Anthropic - I want to convert it to weights for use in massive matrix multiplication operations.

Since the intent is different, the book first sale doctrine applies here to grant transformative. Which is but just one pillar in their defense.

Same goes for what Google did like a decade ago.

• Book - I want to read them.
• Google - I want to convert it to a searchable database of terms that can refer you to the correct ISBN.

And transformative applies too to things like DVDs, it's just under a different framework known as the DMCA.

• DVD - I want to watch the movie.
• MP4 - I want to watch the movie.

That's not transformative.

• DVD - I want to watch the movie.
• MP4 - I want to take clips to make memes.

That is transformative. And again, transformative is but one thing out of the whole "fair use" thing.

This is the thing. Those news articles are oversimplifying a massive amount of legal technicalities. I'm not saying what Anthropic is doing is correct or not. What I am saying is that the conclusion that people are getting from these stories is like the last 10% of a massively technical legal battle. And running with that small amount can lead to conclusions that aren't true.

And I know someone is going to bring it up. No you aren't LEGALLY allows to rip a copy of a DVD for your personal use. Violates the DMCA. BUT in order to sue you, a publisher has to know ahead of time before discovery, that you broke their encryption scheme, which if you only keep it personally, is impossible for them to ever enforce. So we have a law that is written in a way, that if you never put it online, it's impossible to enforce the law. As a summary of how all that works.

Oh! Can I play? Let me transcode a video into a new format and put it on a blu-ray and then delete the old file! Transformative, lol.

You are witnessing the last days of copyright of any form right now - mark my words!

That's why they are scanning them and destroying them. Someone realized they can capitalize knowledge. We're going to get the privilege of letting billionaires gate keep information and knowledge

Unfortunately, that would violate copyright law. In fact, the whole point of the destruction is to circumvent copyright, as apparently there was a ruling that scanning, tokenizing, and then destroying the original is transformative enough to be considered, "fair use."

just remember. the planet is going to be just fine.

remember how quickly wildlife came back when covid shutdowns hit? it'll be like that when the last human dies.

I ask people every now and then, the theme of the answer is similar: one stop shopping, it’s easier than looking elsewhere, answering with the question where else am I going to go? People tend to take the shortest path between A and B.

There are some things you can only find there, sure, but there’s more of a shopping dopamine hit if you look around. Other retailers do exist. Most of the time you can find it elsewhere.

The answer to that question is obvious. They can do it 1000x faster cutting the spin and sending all the pages through a scanner. Otherwise you have to have something flip through every page and take a picture.

Other AI companies can still buy them, they're just rare because nobody bought them in the first place. They destroy them because scanning them is faster and in some jurisdictions that absolves you from copyright infringement since the amount of copies doesn't go up

I've been on Lemmy since 2023 and I've never seen this article. Not everybody is chronically online. Please don't listen to this guy.

They're destroying the books because it's cheaper than scanning them intact.