Leaked emails allegedly reveal Meta torrented terabytes of pirated data
In a recent development, leaked emails have surfaced, allegedly revealing that Meta (formerly known as Facebook) engaged in the torrenting of terabytes of pirated data for training its AI models. These leaked communications have provided new evidence in a copyright case against Meta, filed by book authors who claim that the tech giant illegally used pirated books to train its AI algorithms.
Meta's Use of Pirated Data
The lawsuit against Meta, filed in January, accuses the company of training its AI models on a dataset comprised of pirated ebooks and articles. The unsealed emails indicate that Meta not only used pirated content for training but also admitted to torrenting a significant dataset known as LibGen.
Legal Risks and Internal Concerns
The leaked emails also shed light on Meta employees' awareness of the legal risks associated with their actions. In one email, a Meta research engineer expressed concerns about torrenting from a company laptop, stating that it didn't feel right.
Further internal messages revealed that Meta employees were cautious about the legal implications of their torrenting activities. Despite these concerns, Meta allegedly attempted to conceal its seeding activities and minimize the risk of legal repercussions.
Implications and Reactions
The authors of the lawsuit have condemned Meta's actions, describing the scale of the torrenting scheme as astonishing. They pointed out that even minor instances of data piracy have led to criminal investigations, highlighting the severity of Meta's actions.
As the leaked emails continue to fuel the ongoing legal battle, Meta faces growing scrutiny over its use of pirated content for AI training purposes. The revelations from these communications have reignited the debate on ethical AI practices and intellectual property rights in the tech industry.
For more information, you can refer to the original article on Ars Technica.
