Who is buying old books in Japan for AI training?
Japanese media report a mass acquisition of old books potentially for digitization and use in training AI models.
Japanese media have discovered that unidentified individuals are mass-buying old books in Japanese and sending them abroad. The purchasing scheme looks unusual: buyers are not limited to one topic, taking everything from history and philosophy to law and local studies.
Some antiquarian bookstores sell more than 100 books a day. Since August, over 50 tons of Japanese books have been shipped to the US, equivalent to approximately 100,000 copies.
Despite the lack of direct evidence, the industry suspects the goal is digitizing rare texts for AI training. This has happened before: previously, Anthropic mass-purchased paper books, cut off spines, and scanned pages to create datasets.
As publicly available data on the internet becomes insufficient, old and un-digitized editions are becoming a new valuable resource for generative AI.
Why it matters
AnalysisThis demonstrates that AI developers are turning to rare offline data due to the exhaustion of public internet resources, which may impact the quality and uniqueness of future models.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
How to Build a Custom Design System in Claude Design: From Moodboard to Animated Mascot
Complete practical guide to building a brand design system in Claude Design: visual references, 4 diverging concepts, tokenization, stress-testing, and animated SVG mascot design.
Claude Code Projects: How a Single AI Orchestrates a Team of Autonomous Agents
A deep dive into Claude Code Projects: Coordinator and Threads architecture, parallel branches, shared memory via MEMORY.md, and cloud agent management.
Computer Use via AI
Anthropic's technology (introduced in Claude 3.5 Sonnet) enables neural networks to control operating systems like a human: taking screenshots, analyzing window coordinates, moving the mouse cursor, clicking buttons, and entering text without using specialized APIs.
MCP (Model Context Protocol)
An open standard from Anthropic based on JSON-RPC 2.0 for unified bidirectional connection of AI assistants to external tools, databases, and system environments.