Tit-for-Tat: Porn Producers Counter Meta's "Personal Use" Piracy Defense (torrentfreak.com)

Meta is fighting a copyright suit from adult-content companies Strike 3 Holdings and Counterlife Media that alleges the company used pirated adult films—obtained via BitTorrent—to train AI models. Meta moved to dismiss, arguing the plaintiffs’ IP‐address links are inconclusive, many alleged downloads predate its AI work, and the low average download rate (about 22 per year across dozens of IPs) is consistent with employees or visitors downloading for “personal use.” Plaintiffs rebut with pattern analysis they say shows centralized, non‑human behavior: coordinated downloads from corporate, residential and “hidden” IPs of many different files (multiple versions of Microsoft Office within hours; disparate items tied to the word “origin”), which they argue looks like algorithmic scraping rather than casual user activity. The dispute matters for AI/ML because it attacks how large datasets are assembled and tested methods for attribution. Key technical points include the evidentiary weight of IP-address logs, behavioral signatures that may indicate automated crawling versus human downloads, and the plaintiffs’ novel claim that Meta used popular torrents as “currency” in BitTorrent’s tit‑for‑tat swarms to sustain broader downloads—an argument limited by BitTorrent’s swarm‑specific reciprocal mechanics. The court will decide whether these allegations survive Meta’s motion to dismiss; a hearing is set for January 21, 2026, with potential implications for dataset provenance, corporate liability, and defenses based on “personal use.”