Mass downloading and vector conversion of copyrighted text, imagery, and code for analytical ingestion where the end-user query output is non-substitutive.
Building on the landmark Second Circuit precedent in Authors Guild v. Google, the non-expressive use doctrine establishes that copying works to extract non-expressive analytical, statistical, or functional data constitutes highly transformative fair use under 17 U.S.C. § 107(1). However, that safe harbor assumed the ultimate output was an index or snippet that directed users to the original work. In generative AI, where the model output can substitute for the training work in the commercial market, Factor 4 market harm is fiercely litigated.
While search engine indexing and anti-plagiarism scanning are settled fair use safe harbors, their application to commercial generative foundation models that synthesize competitive creative content remains an active battleground.
