Judge Orders Meta to Name Employees Who Approved Torrenting (torrentfreak.com)

A recent ruling by Magistrate Judge Thomas Hixson requires Meta to disclose the identities of employees involved in its controversial torrenting of pirated books for its AI training datasets. This follows a series of lawsuits about Meta's practices surrounding shadow libraries, with the court previously ruling that training models on copyrighted books constituted fair use. However, the focus now shifts to Meta's use of BitTorrent, with Entrepreneur Media demanding to know who authorized these actions, as they claim Meta contributed to piracy. The significance of this case within the AI/ML community lies in its implications for data acquisition ethics and legal precedents surrounding the use of copyrighted material. Meta has argued that decisions around torrenting were technical and collaborative, meaning no formal approvals exist from higher-ups. However, the court emphasized that all individuals involved in the decision-making process must be identified. This ruling highlights the complexities of AI training data sources, challenging tech companies to navigate legal frameworks while innovating responsibly. The outcome may set important precedents for how AI entities source training data, especially in relation to copyright laws.