News Meta staff torrented nearly 82TB of pirated books for AI training — court records reveal copyright violations

[Admin]

Court records reveal Meta employees were concerned about using torrented data to help train LLaMA.

Meta staff torrented nearly 82TB of pirated books for AI training — court records reveal copyright violations : Read more

 

[Gururu]

Of course they did. And of course they will get away with it.

 

[Shiznizzle]

Did they think they could get away with it?

Yes, they did. And they will get away with it as the fine will be the cost of doing business and not even be a fraction of %1 of one hours worth of profits.

That is if they get fined to begin with. They will laugh at the consequences.

Not in million years will they face the full penalty for each individual infraction. That would shut them down.

My guess is the government will make a deal with them. Share the data and we will slap you o the wrist.

Will they go after googles' Youtube with the same fervor?

 

[COLGeek]

Aside from this particular case, one of the massive complaints regarding LLMs is the siphoning of copyrighted content. One consequence of this has been the pay-walling of many news sources, further diluting the gene pool of actual news vice algorithm enhanced garbage.

Consumers need to be aware of what they are actually consuming.

 

[Pierce2623]

This is the behavior you guys are fundingwith your constant use of “IG”. Bravo.

 

[SomeoneElse23]

These giants make their money largely from advertising.

Block all ads, and they wither away to nothingness.

 

[gg83]

Non of this ai crap is remotely profitable. They had to pirate the stuff, probably hoping they'd settle for much less than if they did it ethically.

 

[P.Amini]

I am angry and mad. Even if they pay full prices it is unfair and inappropriate use of what they consume, let alone pirating it. I am really angry and mad about this.

 

[alrighty_then]

Making a career as an author is looking more dicey each year.

 

[atomicWAR]

This is just stupid. Meta does this and at best they'll get a small fine that doesn't begin to cover each copyrighted piece of content. If the average Joe got hit with this they would end up with a criminal record and a fine (per copyrighted item) that would likely exceed that of Meta's (assuming they get fined at all).

I am not a fan of AI and this certainly doesn't help things. Writers/publishers deserve proper payment for their works. Heck I may use ad-block much of the time for security reasons but even then I make a point of supporting the sites or author's I read on the regular by using affiliate links to buy things or purchasing swag directly from said sources to support them. This just disgusts me. I hope the employees involved are punished and Meta ends up with a fine big enough to truly deter them from such behavior in the future.

 

[ekio]

lol, because you think "Open"AI and Google didn't do the same ??
Like AI knows what happened in any book just by learning their crappy wikipedia page ?

 

[monkeyboz]

Now they are completely disemboding the truth. They had enough people in Facebook and Google to scan an entire library with basic checkout processes and available usage of book scanners. Very scary that they somehow had employees doing so for the project gutenburg but no answers on cross referencing or inclusion of these "scanning" instances other than that the entire archive and server(s) made it onto the torrent network.

Very weird, indeed.

 

[monkeyboz]

Now they are completely disemboding the truth. They had enough people in Facebook and Google to scan an entire library with basic checkout processes and available usage of book scanners. Very scary that they somehow had employees doing so for the project gutenburg but no answers on cross referencing or inclusion of these "scanning" instances other than that the entire archive and server(s) made it onto the torrent network.

Very weird, indeed.

It's even scarier that no one's family members were even responded about when fake recruitment companies placed them inside the Facebook headquarters to further misuse them with the statement that it will be Facebooks fault or the misusage of the word torrent to hide something.

Very scary, some people's families are still unaware about how their family members got stuck inside human trafficking influences as if it were a publisher's interest ... they really need to be careful, the repositories even said they were forced to use Facebook directly in previous court proceedings to keep their human counterparts safe from erasing and re-recording instances.

BTW, some publishers have insisted on a database of their works both released and pre-released outside of Amazon for years, for the safety of the authors time and safety, because some authors have been removed from their works directly in a continued "performance" against their actual processes to archive hours contributed to their actual work.

 

[thestryker]

This is how every large model has been created and the only thing that will stop the behavior is complete destruction of any model created using copyrighted works. I highly doubt this is what the outcome will be so all of the larger companies will just continue on business as usual because no fine will be high enough to deter.

 

[Paul Connell]

Any company that does something like this should loose ALL it's copyright protections for some length of time, maybe starting with a year. During that year, it should be a literal free-for-all, where anyone and everyone gets full access to any of that company's IP they can get their hands on, to use however they see fit, free of charge.

I'm just so sick and tired of corporations guarding their own IP like supermax prison guards, and then straight up stealing other people's IP.

 

[Paul Connell]

I highly doubt this is what the outcome will be so all of the larger companies will just continue on business as usual because no fine will be high enough to deter.

Yep. Not that it needed more clarification, but the last 4 years really put the farce of no one being above the law to rest for good. Justice doesn't exist, it's a fairy tale told to the masses to keep them from revolting.

 

[USAFRet]

This is starting to veer into the Political world.
Let's not do that.

 

[cpkennit83]

Dont put this <Mod Edit> in writing, its like corporate 101.

 

[dalauder]

These giants make their money largely from advertising.

Block all ads, and they wither away to nothingness.

You're not wrong. But I don't quite get where the ad money actually comes from. Most of the ads are to click-bait sites that want to trick you into clicking ads for click-bait. 90% of ads are ads for selling ads, like the ones that were at the bottom of this article on Tom's Hardware.

It's a weird circular reference of bad ad data that might actually be almost totally wasted spending. There are real ads for real products, like you see on YouTube. There are a few reasonable ads for Temu or another online retailer that show me something I already was looking at. But nothing else is even to a place I can spend money!

 

[dalauder]

This is starting to veer into the Political world.
Let's not do that.

Well it started with a court case...which some say is inherently political. But you're still correct.

It just seems like the AI could've been trained in bulk with public domain literature and old National Geographic type stuff, then tuned/pruned with selected data sets. It's like Meta really did it to flaunt that they're too big to get in trouble.

 

[USAFRet]

But I don't quite get where the ad money actually comes from. Most of the ads are to click-bait sites that want to trick you into clicking ads for click-bait. 90% of ads are ads for selling ads, like the ones that were at the bottom of this article on Tom's Hardware.

Ads in front of eyeballs, either TV, internet, or print, are FAR more effective than people think.

They aren't there to make you buy something, but rather to inform you about a product, and to give you a warm fuzzy about the company.
So that the next time you are looking to buy <something> you might think a little better of Company X vs Company Y.

 

[Alwaysmore]

These giants make their money largely from advertising.

Block all ads, and they wither away to nothingness.

They make their money selling your data to the highest bidder. They would still exist if you blocked their ads. They would wither if you deactivated your accounts.

 

[P.Amini]

Ads in front of eyeballs, either TV, internet, or print, are FAR more effective than people think.

They aren't there to make you buy something, but rather to inform you about a product, and to give you a warm fuzzy about the company.
So that the next time you are looking to buy <something> you might think a little better of Company X vs Company Y.

It works the opposite way on me, I try to stay away from advertised products as much as possible. It was the last century when someone found out to be successful at business / selling your products you don't have to raise the quality which itself costs a lot, you just try to advertise it via different methods which costs only a fraction and you'll sell even more this way.

 

[USAFRet]

It works the opposite way on me, I try to stay away from advertised products as much as possible.

Everyone says that, but they're mostly wrong.

The parts in your current PC....how did you know about them, and the company you bought them from?

"inform"

 

[bit_user]

These giants make their money largely from advertising.

Block all ads, and they wither away to nothingness.

It's true. Meta is profiting off you, even if you don't use any of their sites, because they serve ads that show up on the broader web. Setup IP filters to block their ad servers. I think the same is true of Twitter/X, which I've blocked as well.

And, of course, all ad networks track you and collect information about you.