Domain: Intellectual Property Law / Technology Litigation
Persona: Senior Intellectual Property Litigator and Legal Analyst
Tone/Vocabulary: Formal, analytical, objective, and precise.
2. Abstract and Summary
Abstract:
This report analyzes a pivotal development in the ongoing copyright litigation between Meta and a class of authors (including Sarah Silverman and Richard Kadrey). While Meta previously secured a favorable ruling establishing that the use of copyrighted material for Large Language Model (LLM) training constitutes "fair use," the company remains facing claims of direct copyright infringement stemming from the use of BitTorrent to acquire these datasets. Meta’s latest supplemental interrogatory response introduces a novel defense: "fair use by technical necessity." Meta argues that the simultaneous uploading (seeding) inherent to the BitTorrent protocol is an involuntary and essential component of acquiring large-scale datasets for transformative purposes. Furthermore, Meta asserts that the lack of infringing output or demonstrable market harm, coupled with the national interest in maintaining U.S. global leadership in AI, reinforces the validity of its fair use defense.
Litigation Analysis: Meta’s "Technical Necessity" Defense in BitTorrent Infringement Claims
Initial Legal Victory and Residual Risk: Meta previously established that utilizing pirated books to train its Llama LLM is a transformative fair use. However, this did not absolve the company of direct infringement claims related to the specific method of data acquisition—downloading and simultaneously uploading (seeding) files via BitTorrent from "shadow libraries" like Anna’s Archive.
The "Technical Necessity" Defense: In a supplemental interrogatory response, Meta argues that BitTorrent seeding qualifies as fair use. The defense posits that since the BitTorrent protocol automatically uploads data as a function of downloading, the act of "making works available" is an involuntary technical requirement rather than a deliberate choice.
Part-and-Parcel Doctrine: Meta’s counsel contends that because the datasets were only available in bulk via torrents, the distribution of fragments during the download process is "part-and-parcel" of the overarching transformative purpose of AI training.
Authors’ Procedural Challenge: Counsel for the plaintiffs filed a letter with Judge Vince Chhabria on Monday morning, characterizing Meta’s new defense as an "improper end-run" around discovery deadlines. They argue Meta failed to assert this defense during earlier stages of the proceedings despite multiple opportunities.
Meta’s Rebuttal on Timing: Meta’s legal team countered that the defense was explicitly flagged in a joint case management statement in December 2025 (as cited in the text), arguing that the plaintiffs were well aware of the intent to use a fair use defense against the uploading claims.
Admission of No Market Harm: Meta cites deposition testimony where the named authors, including Sarah Silverman, admit they cannot identify any model output that replicates their copyrighted content. Meta argues these admissions negate the "market harm" factor of the fair use four-factor test.
Geopolitical Interest Argument: Meta adds a policy-level layer to its defense, suggesting that the data acquisition was necessary to establish U.S. global leadership in AI, implying that the public benefit of technological supremacy outweighs the technical infringement of distribution.
Key Takeaway: Precedent for AI Sourcing: The ruling by Judge Chhabria on whether to allow the "fair use by technical necessity" defense will set a significant legal precedent for how AI developers can legally source massive datasets from non-traditional or decentralized networks.