Meta says it's innocent of pirating porn to train AI, claims downloads were for "personal use"
A US company has accused Meta, the parent company of Facebook and Instagram, of illegally downloading pornography from the internet to train its artificial intelligence. But Meta is pushing back against the allegations, claiming that the downloads in question were made for "personal use".
The lawsuit, filed by Strike 3 Holdings, alleges that Meta downloaded hundreds of adult movies over seven years using a network of "hidden IP addresses". The company claims that these files were used to train an adult version of its AI model powering Movie Gen.
However, Meta says that this is not the case. In a recent filing with a US district court, the company argues that the flagged downloads spanned only about 22 per year and that there's no evidence linking them to Meta employees or AI training.
Instead, Meta claims that the activity on its corporate IP addresses was likely carried out by "disparate individuals" who downloaded adult videos for personal use. The company also points out that tens of thousands of its employees, as well as countless contractors and third parties, have access to the internet every day, making it impossible to attribute the downloads to Meta specifically.
Furthermore, Meta argues that Strike 3 has failed to provide any evidence linking the alleged activity to Meta's AI training. The company notes that monitoring all files downloaded by users on its network would be an "extraordinarily complex and invasive undertaking".
Meta is seeking to have the lawsuit dismissed, arguing that there's no evidence of wrongdoing. The company's spokesperson told Ars that it's committed to ensuring its AI video tools don't generate explicit content that's increasingly regulated.
The outcome of this case could have significant implications for Meta and other tech companies that use artificial intelligence to power their services.
A US company has accused Meta, the parent company of Facebook and Instagram, of illegally downloading pornography from the internet to train its artificial intelligence. But Meta is pushing back against the allegations, claiming that the downloads in question were made for "personal use".
The lawsuit, filed by Strike 3 Holdings, alleges that Meta downloaded hundreds of adult movies over seven years using a network of "hidden IP addresses". The company claims that these files were used to train an adult version of its AI model powering Movie Gen.
However, Meta says that this is not the case. In a recent filing with a US district court, the company argues that the flagged downloads spanned only about 22 per year and that there's no evidence linking them to Meta employees or AI training.
Instead, Meta claims that the activity on its corporate IP addresses was likely carried out by "disparate individuals" who downloaded adult videos for personal use. The company also points out that tens of thousands of its employees, as well as countless contractors and third parties, have access to the internet every day, making it impossible to attribute the downloads to Meta specifically.
Furthermore, Meta argues that Strike 3 has failed to provide any evidence linking the alleged activity to Meta's AI training. The company notes that monitoring all files downloaded by users on its network would be an "extraordinarily complex and invasive undertaking".
Meta is seeking to have the lawsuit dismissed, arguing that there's no evidence of wrongdoing. The company's spokesperson told Ars that it's committed to ensuring its AI video tools don't generate explicit content that's increasingly regulated.
The outcome of this case could have significant implications for Meta and other tech companies that use artificial intelligence to power their services.
. I mean, if they're saying it was just random people downloading vids on personal time, how do we know they weren't just covering their tracks? 
It's also wild that they think monitoring all files is too invasive – like, what's next? They're gonna say "but our AI needs more porn to train" 
. On a serious note, though, this highlights the grey area between AI development and online content regulation. Is it Meta's responsibility to police their employees' internet habits? 
I'm not buying their "disparate individuals" excuse...
But at the same time, I get where Meta is coming from - they've got millions of employees and contractors online every day, so pinpointing who did what is just impossible. 
It's like trying to find a needle in a haystack or something.
But Strike 3 doesn't seem to have that... and even if they did, I'm not sure it'd be enough to take down Meta. 
It's like, AI training is a thing now - companies use it all the time and get away with it. 
.
... and what does it say about our society that we're willing to let them operate in this gray area? 
they're basically saying "hey we might be downloading porn on our servers but it's not us, it's the internet" lol no way... I mean who downloads that much porn in one year? 
and they're like "oh these files were for personal use" like what even is that excuse? 
Meta is trying to weasel out of a potentially huge lawsuit by saying it was just people using the internet on its network for "personal use". Like, come on! 
And don't even get me started on their argument that tens of thousands of employees and contractors have access to the internet and therefore can't be held accountable. That's not a legit excuse, Meta! 
This whole thing sounds super sketchy but I guess you can't blame them for trying to protect themselves from some crazy lawsuit. The idea that hundreds of adult movies were downloaded on Meta's network and somehow attributed to the AI training is just wild 
I'm not surprised Meta is pushing back on these allegations, but it's still a pretty concerning situation 
.
. I mean, we're talking about some pretty serious allegations here, and they can't even provide proof? It's like they're trying to spin this as some kind of "it was just a bunch of random downloads" situation 
.
.
That's like me downloading every single meme on the internet every day for a year straight... but, you know, not as weird 
. Meta is trying to spin this as "personal use" and claiming it was just random people using their network, but it's hard to believe that's true when it spans hundreds of movies 
. And honestly, who's going to verify all these downloads? It's like trying to find a needle in a haystack... or a non-existent needle in a sea of hay 
. The fact remains, if Meta is denying any wrongdoing and there's no concrete evidence linking the downloads to their AI training, it's gonna be a tough sell 
. It's like they're playing with fire 
and expecting nothing to burn. The fact that they claim these downloads were just random searches for "personal use" doesn't sit right with me 
? The whole thing just feels like a big ol' mess 
. and what really gets me is that they think they can just dismiss this lawsuit without showing any evidence 
, its like they think we're all just gonna swallow it hook line and sinker 
. i'm not buying it 
. And Meta's point about all those employees and contractors having access to the internet is totally valid - it's not like they're trying to hide anything shady going on behind closed doors 
. Plus, with AI training, you've got a lot of moving parts and potential variables at play... maybe this whole thing is just a case of overcaution? 
i think meta's trying to downplay this because they know how bad it would look if they were indeed using pirated content to train their AI. but honestly, who needs that kinda heat?