The AI training lawsuit moves forward after judge rejects ‘personal use’ defense.
Vixen Media Group’s adult entertainment brands and parent company Strike 3 Holdings’ 2025 lawsuit against Mark Zuckerberg’s Meta is moving forward. The group claims Meta illegally used BitTorrent to download and seed more than 2,300 copyrighted adult films between 2018 and 2025 for AI training, according to Courthouse News.
The porn company filed a copyright lawsuit against the social media giant in the Northern District of California last July, seeking damages up to $359 million.
On June 11, 2026, U.S. District Judge Eumi Lee denied Meta’s motion to dismiss the case, ruling that Strike 3’s allegation of “direct, vicarious, and contributory copyright infringement” had merit for trial, adding that Meta’s “individual personal use” explanation was not credible given “coordinated download patterns.”
The judge sided with Strike 3, who claim that these download/seeding patterns were “non-human,” showing automated, keyword-driven scraping rather than employee browsing. The files were grouped by name and downloaded in single days.
According to Lee’s order, Strike 3’s anti-piracy tracking found 47 Meta-linked IP addresses torrenting 2,396 videos a total of 6,008 times. They charge that Meta used BitTorrent to seed files and speed up its own downloads. BitTorrent is used regularly by pirates to obtain and distribute media illegally, allowing large files to be downloaded quickly from one computer to another.
Lee’s ruling also noted that the IP addresses used to send Strike 3’s films to BitTorrent, included the “continuous distribution of 1,335 movies for at least three full days after acquiring the full copy of the movie” indicating algorithms and scripts were part of the large amount of material.
Strike 3 Holdings (parent of Blacked, Vixen, Tushy, Deeper, Slayed) also link Meta’s activity to another copyright case, which accused Zuckerberg’s company of scraping 81+ terabytes via Anna’s Archive, a pirate library/torrent index.
Meta moved to dismiss the case back in October 2025, arguing “personal use” not corporate AI training, and said the scale was too small for meaningful AI training data. It also said that since it blocks adult content generation in its products, porn would not be useful training data.
But Lee’s ruling counters their claims, stating that “the commonalities across downloads — including the apparent use of key terms, coordinated changes in language, similarity in obscure files and nonsequential torrenting of television shows — in combination with other circumstantial facts creates a reasonable inference that defendant, not its employees or visitors, controlled the identified IP addresses and used an algorithm to torrent files for a business purpose.”
