Meta Accused of Using Pirated Porn to Train Its AI Superintelligence
▼ Summary
– Strike 3 Holdings is suing Meta for allegedly torrenting and seeding its copyright-protected adult videos since 2018 to train AI models.
– The company claims Meta sought its content for unique visual angles and scenes to gain a competitive edge in AI fluidity and realism.
– The lawsuit alleges Meta illegally distributed 2,396 videos via BitTorrent, making them accessible to minors without age verification.
– Exhibits list titles from mainstream shows and other adult content, including potentially concerning material involving young actors and weapons.
– Strike 3 is seeking $350 million in statutory penalties, while Meta disputes the claims and is reviewing the complaint.
A major legal battle has erupted over the alleged use of copyrighted adult content to train artificial intelligence systems. Strike 3 Holdings, a producer of what it describes as ethical and feminist adult films, has filed a federal lawsuit against Meta, accusing the tech giant of using pirated material to advance its AI development. The complaint, initially filed in July and recently unsealed, contends that Meta has been systematically downloading and distributing Strike 3’s videos via BitTorrent since 2018.
According to the suit, Meta sought out this specific type of content to gain access to visual perspectives, anatomical details, and extended scenes that are seldom available in mainstream media. These elements are considered valuable for creating more realistic and fluid AI-generated visuals. An attorney for Strike 3, Christian Waugh, emphasized that the company’s content offers a competitive advantage in achieving what Mark Zuckerberg has termed AI “superintelligence.”
The complaint goes further, alleging that Meta illegally downloaded and shared 2,396 copyrighted videos belonging to Strike 3. Because BitTorrent lacks age verification, the suit claims this activity made pornographic material accessible to minors. Meta is accused of using these videos as “currency” to facilitate the downloading of other content necessary for AI training.
Exhibits attached to the lawsuit include a wide array of media allegedly sourced by Meta, ranging from popular TV series like Yellowstone and South Park to more concerning categories. The list features adult content involving very young performers, along with titles related to firearms and even ironic inclusions like Intellectual Property Rights in Cyberspace.
Legal experts warn that using adult content in AI training sets carries significant reputational risks. Matthew Sag, a law professor specializing in AI and data science, illustrated the danger with a hypothetical scenario: a student asking an AI for a pizza delivery video and unintentionally receiving pornographic results.
Strike 3 claims it identified the infringements through proprietary detection systems, tracing the activity to 47 IP addresses linked to Meta. The company is seeking $350 million in statutory penalties.
A Meta spokesperson, Christopher Sgro, responded by stating that the company is reviewing the complaint but disputes the accuracy of the allegations. This case emerges amid Meta’s ambitious AI initiatives, including the recent release of its V-JEPA 2 “world model,” trained on one million hours of unspecified “internet video.” Zuckerberg has repeatedly emphasized his goal of embedding “personal superintelligence” into consumer products like Meta’s smart glasses, raising questions about the boundaries of data sourcing in the race for AI dominance.
(Source: Wired)
