A lawsuit filed in California federal court is turning a spotlight on how tech giants gather raw material for artificial intelligence tools.
Strike 3 Holdings accuses Meta of downloading thousands of its copyrighted adult films using BitTorrent without permission, then repurposing that material to train its advanced AI systems.
Court records unsealed last week outline the company’s claim that Meta targeted these videos for their unique visual perspectives and long, uninterrupted scenes, which often are not available in popular TV or cinema.
Christian Waugh, attorney for Strike 3, argued Meta wanted these explicit works “because it can give them a competitive advantage for the quality, fluidity, and humanity of the AI.” According to the complaint, Meta allegedly accessed 2,396 of Strike 3’s adult videos and swapped those files online to create a pool of content for AI model training.
Concerns About Content and Potential Risks
Strike 3 says Meta’s actions could have exposed minors to explicit content, as BitTorrent platforms do not include any age verification measures.
The list of alleged material is wide-ranging, pulled both from well-known shows like Yellowstone and South Park and from explicit sites with names that suggest actors may be quite young. There are even segments dealing with weapons, political extremism, and copyright itself.
Using explicit adult entertainment as AI training material invites trouble for major tech companies, says Matthew Sag, a law professor at Emory University who specializes in artificial intelligence. He pointed out the risk public-facing AI could inadvertently serve up adult content to unsuspecting users, with potentially disastrous results. “Imagine a middle school student asks a Meta AI model for a video about pizza delivery, and before you know it, it’s porn,” Sag cautioned.
Strike 3 says it operates its own detection technology to track where and how its videos appear, allowing it to find connections to forty-seven Meta-associated network addresses. The studio is now seeking damages of three hundred fifty million dollars for these supposed violations.
Meta’s spokesperson, Christopher Sgro, commented, “We’re reviewing the complaint, but we don’t believe Strike’s claims are accurate.”
Documents tied to the case say Meta’s research team trained its recent V-JEPA 2 AI release on a million hours of “internet video,” without clarifying exactly what sources were used. Meta’s CEO, Mark Zuckerberg, has championed building “superintelligent” systems, with the ambition that these tools could one day personalize technology in meaningful ways, like through Meta’s new smart glasses.
