The copyright fight between Strike 3 Holdings and Meta got more specific this week, and the detail at its center is a residential internet connection. Strike 3, along with Counterlife Media, the parent of adult brands Vixen and Blacked, is suing Meta in the Northern District of California, and the amended complaint now covers 2,973 films with statutory damages reaching roughly $446 million. Meta filed its formal Answer during the week of July 20, denying wrongdoing and laying out how it plans to fight.
The allegation is heavy: that pirated adult films moved through a pipeline connected to Meta’s AI training. The Answer is Meta’s attempt to sever that connection at its weakest link.
Meta tied the Comcast IP address behind 97 of the allegedly infringed works to a specific household, and named the person: the adult son of a Meta employee. Per the filing, that individual worked as a contingent worker at Meta from 2022 to 2024 and later as a data engineer from 2025 to 2026. Meta’s argument is that torrenting on a family member’s home connection is personal use, unrelated to the company’s corporate AI work.
Around that, Meta stacked several defenses. It raised a de minimis argument, claiming that partial BitTorrent file fragments do not add up to a usable copy. It reserved a fair-use defense without asserting it yet, mirroring its posture in the related Kadrey case over AI training. It invoked the statute of limitations and a Fifth Amendment challenge to the size of statutory damages. And it disputed whether Strike 3’s films are properly copyrighted original works in the first place.
Meta separately acknowledged downloading portions of publicly available text datasets to train its LLaMA models, while denying those datasets included adult video.
Why this one matters
Most piracy suits are about lost sales. This one is about training data, which is why the industry should track it past the salacious headline. If a court accepts that fragmentary BitTorrent transfers are de minimis, or that a company is not responsible for what flows through an employee’s home connection, those findings ripple into every case where a rightsholder alleges an AI firm ingested pirated work.
The procedural clock is running. Meta’s motion to dismiss was denied in June, the amended complaint was entered in early July, a mediation deadline sits in early August, and trial is set for February 2028. That timeline means the case has room to become a reference point rather than a footnote. For an industry whose content is among the most pirated on the internet, and now among the most valuable as training data, the question of who is liable when that content is scraped is not academic. It is the whole ballgame.
