By asahifile Meta Denies Torrenting Porn to Train AI, Claims Downloads Were for “Personal Use”
In a recent court filing, Meta has responded to allegations claiming the company illegally pirated pornographic content to train its artificial intelligence models. The tech giant’s motion to dismiss the lawsuit was filed after Strike 3 Holdings discovered illegal downloads of several adult films associated with Meta’s corporate IP addresses, raising concerns that these downloads were conducted using a covert network of hidden IPs.
The lawsuit, which could potentially demand damages exceeding $350 million, accuses Meta of illicitly downloading pornographic material to secretly train an unpublicized adult-themed version of their AI model, known as Movie Gen. However, Meta has refuted these claims, labeling them as baseless and suggesting that Strike 3 is a “copyright troll” that files frivolous lawsuits for profit.
In its response, Meta argues that there is no evidence indicating the corporation directed or was aware of the downloads, which hit approximately 2,400 adult movies owned by Strike 3. They maintained that Strike 3 failed to substantiate its claims, stating, “these claims are bogus,” as representatives of the company insist that the flagged downloads dating back to 2018 were for personal use only.
Meta supports its defense by stating that the downloads were isolated incidents, averaging about 22 instances per year, and likely conducted by individuals tied to the company for personal rather than professional motivations. Furthermore, the company asserts that the allegations lack specifics, such as identifying the individuals supposedly behind the downloads or linking any employee to the act of downloading adult content.
Amidst these accusations, Meta stresses their commitment to ensuring that their AI systems do not generate explicit or adult content, emphasizing their adherence to a prohibition against using such material in AI training. In its filing, Meta argues, “We don’t want this type of content, and we take deliberate steps to avoid training our AI on such materials.”
The issue of how legal downloads are monitored also arises, with Meta contending that the complexity of tracking every file downloaded through its network would be an unreasonable expectation. The company has expressed confidence that the court will uphold its dismissal motion, asserting that Strike 3 has not proven any correlation between Meta and the alleged illegal downloads.
As the case develops, both parties prepare for further legal proceedings, with Strike 3 having two weeks to respond to Meta’s objections. The outcome could have implications for Meta’s reputation and its operational practices regarding data sourcing for AI training.
This summary captures the essence of the original article while maintaining clarity and coherence, suitable for an audience seeking an overview of the legal dispute involving Meta and Strike 3 Holdings.