Judge Lets $359 Million Porn Copyright Suit Against Meta Proceed Over 2,300 AI Training Downloads
Updated
Updated · Mashable · Jun 15
Judge Lets $359 Million Porn Copyright Suit Against Meta Proceed Over 2,300 AI Training Downloads
3 articles · Updated · Mashable · Jun 15
Summary
June 11's order found Strike 3 Holdings and Counterlife Media plausibly alleged Meta directly, vicariously and contributorily infringed copyrights by torrenting adult films, allowing the case to move forward.
More than 2,300 movies were allegedly downloaded between 2018 and 2025 through BitTorrent to train Meta's AI models, with IP addresses tied to Meta offices showing mass, non-human download patterns.
Meta had called the claims nonsensical and said the downloads were for personal use, but Judge Eumi K. Lee said same-day torrenting of similarly named files across genres strained credulity as coincidence.
The plaintiffs say they identified Meta's BitTorrent activity after discovery in a separate 2025 AI-books piracy case; Meta won that case in June 2025, though the judge there signaled different legal arguments might succeed.
After a rival's $1.5B settlement, is Meta's AI built on a ticking time bomb of pirated data?
While banning others from scraping its data, did Meta secretly build its AI with pirated videos and books?
Meta Sued for Allegedly Downloading 81.7TB of Adult Films to Train AI: Copyright and Industry Implications
Overview
A California federal judge has ruled that Meta must face a copyright lawsuit from Strike 3 Holdings, which claims Meta downloaded its adult films to train generative AI models. The case centers on allegations of a coordinated effort by Meta to gather copyrighted content for AI training, echoing earlier accusations from book authors about large-scale data torrenting. This lawsuit is seen as a potential turning point, not only for Strike 3’s legal strategy but also as a precedent for other copyright owners challenging AI companies over data use. The outcome could reshape how AI firms acquire and use training data.