Why This Topic Landed in Japan

A report by Nippon TV's news network said online used-bookstores across Japan have seen a wave of mysterious bulk orders since around August — some shops reporting "100 books sold a day" or sales quintupling — and that follow-up reporting found records of over 50 tons of used books exported to the US. The industry's suspicion is that the buyer is collecting training data for generative AI; who the buyers are and whether the purpose is really AI training have not been independently confirmed. What made it explode online is the image of books being shredded and scanned, then discarded, with no return to the authors or publishers — read by many as free-riding on Japanese culture. At the same time, Japan's current copyright law (Article 30-4) broadly permits information analysis for AI development, so "it's legal, but it doesn't sit right emotionally" became a common refrain. A US case in which AI company Anthropic was said to have scanned books at scale is cited as a precedent, though such details remain unverified. The story also fed a familiar lament: that Japan cannot put its own knowledge to work in domestic AI.

Key Reaction Themes

  • Alarm over IP outflow and free-riding on culture — The largest share framed it as valuable documents "taken overseas" with nothing returned to their creators, some adding that the sting comes specifically from a foreign country doing it rather than Japanese firms or the government.
  • "It's legal under copyright law" — A substantial group pointed out that data analysis for AI is permitted in Japan and that "what a buyer does with what they bought is their business."
  • "Used books get discarded anyway — better preserved as data" — An acceptance view: documents that would otherwise be thrown out and forgotten are at least being used again.
  • Self-mockery over Japan's AI lag — A smaller strand lamented that "it'll all be siphoned off before Japan's own AI even gets going," and that "the buyer may not even be America — it could be China."

What Japanese Netizens Are Saying

Suspected bulk-buying of Japanese used books by US AI firms

A surge of mysterious bulk orders at used-bookstores and records of a 50-ton-plus export to the US were reported as possible data-gathering for generative AI.

ライブドアニュース
ライブドアニュース
@livedoornews
·

【判明】古書店で「謎の大量注文」相次ぐ…AI学習か 米国向け「日本の本 50トン」の輸出記録も news.livedoor.com/article/detail… 日本各地のオンライン古書店に対して、8月ごろから書籍の大量注文が相次いでいる。生成AIの学習用との見方も広がり、日本から米国に50トン以上が輸出された記録もあるという。

Image
Reply
Read 211 replies

Sources: Yahoo! news, 2ch

Comments:

  • "Isn't this copyright infringement?"
  • "Once you've bought them, what you do with them is up to the buyer. Nothing to be done."
  • "The nasty part is that after scanning them, selling or donating them becomes illegal, so they have to shred and scrap every last one."
  • "America just keeps pushing ahead. While we're saying stupid things like this, the gap between us will grow as wide as heaven and earth."
  • "Books are literally lumps of knowledge — of course they should be collected and covered comprehensively."
  • "So archivally valuable material, precious bound editions, actual technical artifacts all vanish like this. Overseas probably has its own brutal reality too."
  • "They're used books nobody buys anyway; if they're destined to become toilet paper eventually, better to feed them to the AI."
  • "They used to say AI would struggle to develop in Japan because there's no big data here — but look, there was big data after all."
  • "These are documents that, even in a used bookstore, would eventually be discarded and forgotten, so if they get attention and are put to use again, the authors would probably be glad. The only thing that feels off is that it's a foreign country doing it, not the Japanese government or Japanese companies. Japanese people should probably take a bit more interest in their own history, culture, and civilization."
  • "Parody would be one thing, but who's happy about unauthorized secondary use of their own work…? It's scary how many people have this weirdly communist, clueless take. Not grasping that using a book and using a book's contents are different things is truly a failure of education."
  • "Outflow of intellectual property and the collapse of the language barrier — what will be left for Japan, I wonder?"
  • "Sorry to burst your bubble, but books keep getting printed endlessly, so buying them is no help at all…? And used books that ordinary people would buy are, at best, only valuable to collectors — they aren't even cultural assets."
  • "Japan should just do this with its own AI."
  • "Taking data from bad books only produces garbage data. Really, good books need to be chosen by human eyes."