Japan's Publishing Industry Outraged Over Suspected Mass Sale of 100,000 Books to U.S. AI Firm Anthropic

Summary A growing international dispute over generative artificial intelligence and intellectual property has hit the East Asian publishing scene. The Japan Book Publishers Association, representing roughly 380 prominent publishers including giants Kodansha and Bungeishunju, has officially demanded answers from a major book distributor suspected of mass-selling books to U.S.-based artificial intelligence developer Anthropic for model training. According to Japanese media reports cited by Yonhap News, court documents filed in an ongoing copyright lawsuit against Anthropic in the United States reportedly referenced book purchases sourced from Nippon Shuppan Hanbai (Nippan), one of Japan's foremost book distributors. In response, the publishers' association issued a formal inquiry to Nippan, asking whether publications by its member publishers were included in the sales and demanding the distributor's stance on exporting vast volumes of books to overseas technology firms without the explicit consent of original publishers and authors. The controversy sheds light on mysterious buying patterns observed recently across Japanese used-book districts, where specialized texts in disciplines such as philosophy, history, medicine, and law began vanishing from shelves in unprecedented quantities. Observers had suspected that AI developers or intermediary suppliers were systematically sweeping up specialized print titles to digitize and feed them into language models before discarding the physical copies. Japanese broadcaster Nippon TV highlighted the sheer scale of the operation, reporting that a subsidiary of a major domestic book distributor shipped over 50 tons of books to the United States—an astronomical haul estimated at approximately 100,000 physical volumes. An official from the Japan Book Publishers Association emphasized that if books were knowingly supplied for artificial intelligence training, it represents an unacceptable bypass of creators' rights, stating firmly that publications should never be sold for AI ingestion without authorization from rights holders. As cross-border AI data harvesting extends into physical international print archives, the clash is drawing significant attention across Asian cultural and publishing industries navigating the delicate balance between technological expansion and creator protections.