Uploading Pirated Books via BitTorrent Qualifies as Fair Use, Meta Argues
Meta is escalating its legal battle against authors, arguing that uploading pirated books via BitTorrent is a necessary 'fair use' action when acquiring data for AI training. This novel defense, dubbed 'fair use by technical necessity,' is presented as an inherent function of the BitTorrent protocol. The ruling will be critical for future AI development, as it challenges traditional copyright infringement notions in the context of large-scale data acquisition for LLMs.
The Lowdown
Meta, the tech giant behind Facebook and Instagram, is embroiled in a significant legal skirmish with authors over the use of copyrighted works to train its artificial intelligence models. While a court previously deemed the training of its Llama LLM with pirated books as fair use, Meta still faces claims of direct copyright infringement for downloading and sharing these books via BitTorrent.
- Meta has now introduced a new defense, asserting that the act of uploading pirated books, which is inherent to the BitTorrent protocol during the download process, also qualifies as fair use.
- The company argues that this uploading was a technical necessity for obtaining bulk datasets from sources like Anna's Archive, which were only practically available through torrent downloads.
- Authors' lawyers have vehemently opposed this new defense, characterizing it as an improper, late filing after discovery deadlines, despite Meta's claims that it was previously flagged.
- Meta further bolsters its stance by citing authors' own depositions, where they admitted to no knowledge of infringing output from Meta's models or market harm, echoing arguments that secured its initial fair use victory for model training.
- Beyond the legal technicalities, Meta frames its investment in AI and the need for comprehensive training data as vital for maintaining U.S. global leadership in artificial intelligence.
The upcoming decision from Judge Vince Chhabria regarding the admissibility of this 'fair use by technical necessity' argument is poised to set a critical precedent for ongoing and future AI-related lawsuits concerning data sourcing from 'shadow libraries.' This ruling could significantly redefine the boundaries of copyright law for AI development.