>Meta, naturally, argues that no! No! That porn was, uh…it was for personal use! Yeah, it was for…me, and not for large-scale copyright infringement!
>[...]
>Meta’s motion to dismiss the case calls Strike 3’s torrent-tracking “guesswork and innuendo” and argues the supposed downloads, which are roughly 22 per year, are too few to have any use in training AI. If anything, the company says, that pattern looks less like corporate malfeasance and more like “private personal use.”
The article seems skeptical of Meta, but their defense seems... fairly reasonable? They have hundreds of thousands of employees, so the prospect of all of them combined accidentally torrenting 22 porn movies per year without a VPN doesn't seem too implausible.
My skepticism would stem from the usual corporate firewall scanning. I routinely get blocked going to innocuous things which are on an uncommon domain or are otherwise on some block list. Recently, I could not visit the DuckLake homepage which is on the .select domain.
That torrents and/or porn are not hard blocked is surprising. For the low volume, I suppose anything can get through.
> argues the supposed downloads, which are roughly 22 per year
the math doesn't add up since that implies it took 100+ years to accumulate 2400 violations, but just looking at 2400 torrents, how many employees does Meta have? With at least several thousand guys, if some have poor 'keeping work and home separate' hygiene, and especially with work-from-home, they probably just forgot to disconnect from VPN after work. Seems more likely than a very sloppy AI training scheme. If they wanted to so such a scheme, it's so easy to do using a public VPN.
EDIT: If I'm reading right, they're actually just basing their suit on HOME IP addresses. So we're being asked to believe the plan was:
1. Top secret porn AI plan is hatched
2. Management considers how to accomplish this and decided the best course of action is to tell dozens of engineers to go home and download a handful of porn torrents each and bring them into work to train it.
This is funny, but from the article it also seems like a pretty frivolous law suit that is more of a shakedown racket than protecting the actual business.
From the article: this amounts to only about 20 videos per year and the evidence is based on home IP addresses of employees.
Such as:
“The father of a Meta contractor whose home IP address allegedly downloaded 97 videos. Strike 3 suggests this links Meta to more infringement. Meta counters that it only proves someone’s dad is super into porn and has no VPN”.
Pretty weak evidence of any malfeasance on Meta’s part.
>From the article: this amounts to only about 20 videos per year and the evidence is based on home IP addresses of employees. Such as: “The father of a Meta contractor whose home IP address allegedly downloaded 97 videos. Strike 3 suggests this links Meta to more infringement. Meta counters that it only proves someone’s dad is super into porn and has no VPN”.
The linked article mentions there was torrenting activity from Meta's ip blocks as well:
>This prompted Strike 3 and Counterlife Media to search for Meta-linked IP addresses in their archive of collected BitTorrent data. This scan revealed that forty-seven IP addresses, identified as owned by Facebook, allegedly infringed their copyrighted works.
> unlike lawsuits raised by book authors whose works are part of an enormous dataset used to train AI, the activity on Meta’s corporate IP addresses only amounted to about 22 downloads per year
It's an entertaining headline, but it actually seems legit rather than a creative legal defense.
I can remember at least one incident when someone on the bus from SF to campus was actually watching porn on their laptop. IDK if they were torrenting it at home or at work, but there were many emails about not streaming content on the shuttle bus WiFi.
> In 2014, the couple had filed a third of all US copyright litigation that year: 1,300 infringement cases.
> According to Westlaw and Pacer data from the past three years, Strike 3 accounted for 50% of the federal copyright docket all on its own.
That implies there are few copyright infringement lawsuits for a country of 340 million people. Do I misunderstand? Does the denominator all copyright infringement enforcement processes? Maybe there are mechanisms besides federal courts, and of course it omits lawsuits threatened but never filed (though Strike 3 files them and then subpeonas IP addresses).
I think your understanding is correct. Lawsuits are usually a last resort after negotiations fail but in these torrent cases they have to file suit to find who to negotiate with.
Use the dirty-net next time the non-meta ASN for doing pentesting. Every company should route non-essential traffic out alternate circuits to keep their employee IP addresses and behavior out of logs that point back to their employer. This does not preclude scrubbing the traffic with ones DLP, MitM proxies, etc... Its just another route. Keep YT and porn off the corporate circuits. Maybe even go so far as to have multiple SNAT pools for different categories of non-work related content. Make the dirty-net the default routes and only route meta-destine networks over the corporate specific networks.
I set up something like this ages ago in a company that was acquired by a company run by a literal mobster. I had a 1U server with two interfaces that routed my coworkers out a path that bypassed the mob monitored devices. To the uppers it just appeared my coworkers were really dedicated and not wasting time on Youtube, LinkedIn, Facebook, etc...
Using torrent is actually an argument in favour of AI usage - people, I think, usually just watch movies on the web? Also, with torrents you would have to wait, while, as a proverb says, "a spoon is needed before the dinner". Or these are the premium type of movies that can be found for free only at torrent sites and are downloaded for later viewing?
> Notably, the flagged downloads spanned seven years, starting in 2018. That’s about four years before Meta’s AI efforts “researching Multimodal Models and Generative Video” began
> Instead, Meta argued, available evidence “is plainly indicative” that the flagged adult content was torrented for “private personal use”—since the small amount linked to Meta IP addresses and employees represented only “a few dozen titles per year intermittently obtained one file at a time.”
Personal use sounds plausible to me when imagining something like 70k employees and a bit of a "bro" culture.
"Only in the world of AI algorithm training can you claim that you were torrenting 2,400 porn videos for personal use and have that seem like the lesser of two evils."
What happened to Vice?
Is it a private equity content mill now? (a la newsweek, ex-gawker properties)
(I'm asking because moralizing about torrenting porn is not a very 2010s vice thing, and I lost track of it)
>Meta, naturally, argues that no! No! That porn was, uh…it was for personal use! Yeah, it was for…me, and not for large-scale copyright infringement!
>[...]
>Meta’s motion to dismiss the case calls Strike 3’s torrent-tracking “guesswork and innuendo” and argues the supposed downloads, which are roughly 22 per year, are too few to have any use in training AI. If anything, the company says, that pattern looks less like corporate malfeasance and more like “private personal use.”
The article seems skeptical of Meta, but their defense seems... fairly reasonable? They have hundreds of thousands of employees, so the prospect of all of them combined accidentally torrenting 22 porn movies per year without a VPN doesn't seem too implausible.
My skepticism would stem from the usual corporate firewall scanning. I routinely get blocked going to innocuous things which are on an uncommon domain or are otherwise on some block list. Recently, I could not visit the DuckLake homepage which is on the .select domain.
That torrents and/or porn are not hard blocked is surprising. For the low volume, I suppose anything can get through.
Does FB not have the equivalent of GoogleGuest SSID which is open to public unauthenticated where reachable?
I've certainly downloaded Ubuntu over torrent on that wifi.
Large tech companies rarely block internet access, since internet access is critical for employees at tech companies.
> argues the supposed downloads, which are roughly 22 per year
the math doesn't add up since that implies it took 100+ years to accumulate 2400 violations, but just looking at 2400 torrents, how many employees does Meta have? With at least several thousand guys, if some have poor 'keeping work and home separate' hygiene, and especially with work-from-home, they probably just forgot to disconnect from VPN after work. Seems more likely than a very sloppy AI training scheme. If they wanted to so such a scheme, it's so easy to do using a public VPN.
EDIT: If I'm reading right, they're actually just basing their suit on HOME IP addresses. So we're being asked to believe the plan was:
1. Top secret porn AI plan is hatched
2. Management considers how to accomplish this and decided the best course of action is to tell dozens of engineers to go home and download a handful of porn torrents each and bring them into work to train it.
3. ???
4. Profit!
> With at least several thousand guys
Over 50,000 employees, per various sources.
This is funny, but from the article it also seems like a pretty frivolous law suit that is more of a shakedown racket than protecting the actual business.
From the article: this amounts to only about 20 videos per year and the evidence is based on home IP addresses of employees. Such as: “The father of a Meta contractor whose home IP address allegedly downloaded 97 videos. Strike 3 suggests this links Meta to more infringement. Meta counters that it only proves someone’s dad is super into porn and has no VPN”.
Pretty weak evidence of any malfeasance on Meta’s part.
>From the article: this amounts to only about 20 videos per year and the evidence is based on home IP addresses of employees. Such as: “The father of a Meta contractor whose home IP address allegedly downloaded 97 videos. Strike 3 suggests this links Meta to more infringement. Meta counters that it only proves someone’s dad is super into porn and has no VPN”.
The linked article mentions there was torrenting activity from Meta's ip blocks as well:
>This prompted Strike 3 and Counterlife Media to search for Meta-linked IP addresses in their archive of collected BitTorrent data. This scan revealed that forty-seven IP addresses, identified as owned by Facebook, allegedly infringed their copyrighted works.
Another article explains that there were more IPs (not officially registered for Meta) and more movies [1]
[1] https://torrentfreak.com/copyright-lawsuit-accuses-meta-of-p...
You gotta hand it to Meta, that's one creative defense. "Sorry your honor, our entire engineering team are certified gooners. We have documentation."
> unlike lawsuits raised by book authors whose works are part of an enormous dataset used to train AI, the activity on Meta’s corporate IP addresses only amounted to about 22 downloads per year
It's an entertaining headline, but it actually seems legit rather than a creative legal defense.
I can remember at least one incident when someone on the bus from SF to campus was actually watching porn on their laptop. IDK if they were torrenting it at home or at work, but there were many emails about not streaming content on the shuttle bus WiFi.
The linked Guardian article about the copyright trolls, Strike 3, has some interesting tidbits about copyright infringement lawsuits:
https://www.theguardian.com/society/ng-interactive/2025/nov/...
> In 2014, the couple had filed a third of all US copyright litigation that year: 1,300 infringement cases.
> According to Westlaw and Pacer data from the past three years, Strike 3 accounted for 50% of the federal copyright docket all on its own.
That implies there are few copyright infringement lawsuits for a country of 340 million people. Do I misunderstand? Does the denominator all copyright infringement enforcement processes? Maybe there are mechanisms besides federal courts, and of course it omits lawsuits threatened but never filed (though Strike 3 files them and then subpeonas IP addresses).
I think your understanding is correct. Lawsuits are usually a last resort after negotiations fail but in these torrent cases they have to file suit to find who to negotiate with.
I actually wouldn't be surprised if it turned out employees downloaded porn on company computers 2400 times at company the size of Meta.
Use the dirty-net next time the non-meta ASN for doing pentesting. Every company should route non-essential traffic out alternate circuits to keep their employee IP addresses and behavior out of logs that point back to their employer. This does not preclude scrubbing the traffic with ones DLP, MitM proxies, etc... Its just another route. Keep YT and porn off the corporate circuits. Maybe even go so far as to have multiple SNAT pools for different categories of non-work related content. Make the dirty-net the default routes and only route meta-destine networks over the corporate specific networks.
I set up something like this ages ago in a company that was acquired by a company run by a literal mobster. I had a 1U server with two interfaces that routed my coworkers out a path that bypassed the mob monitored devices. To the uppers it just appeared my coworkers were really dedicated and not wasting time on Youtube, LinkedIn, Facebook, etc...
Previous:
https://news.ycombinator.com/item?id=45751202
https://news.ycombinator.com/item?id=45937565
Using torrent is actually an argument in favour of AI usage - people, I think, usually just watch movies on the web? Also, with torrents you would have to wait, while, as a proverb says, "a spoon is needed before the dinner". Or these are the premium type of movies that can be found for free only at torrent sites and are downloaded for later viewing?
From the linked Ars article:
> Notably, the flagged downloads spanned seven years, starting in 2018. That’s about four years before Meta’s AI efforts “researching Multimodal Models and Generative Video” began
> Instead, Meta argued, available evidence “is plainly indicative” that the flagged adult content was torrented for “private personal use”—since the small amount linked to Meta IP addresses and employees represented only “a few dozen titles per year intermittently obtained one file at a time.”
Personal use sounds plausible to me when imagining something like 70k employees and a bit of a "bro" culture.
Previously:
https://news.ycombinator.com/item?id=45751202
https://news.ycombinator.com/item?id=45753440
"Only in the world of AI algorithm training can you claim that you were torrenting 2,400 porn videos for personal use and have that seem like the lesser of two evils."
What happened to Vice?
Is it a private equity content mill now? (a la newsweek, ex-gawker properties)
(I'm asking because moralizing about torrenting porn is not a very 2010s vice thing, and I lost track of it)
It’s owned by private equity, per wikipedia.