Secondhand booksellers across the UK and Ireland are reporting an unusual surge in large, random orders from opaque buyers, a pattern that has prompted speculation that AI companies are buying printed material for training data. The Guardian reported that sellers have received requests from buyers in North America, Europe and the UK, often with no obvious subject or collecting theme.
At Barter Books in Alnwick, co-owner Stuart Manley said three buyers had purchased hundreds of unrelated books for about £4,000. One order mixed an Estonian translation of a John le Carré novel, a particular edition of Anne Brontë’s *Agnes Grey* and a 1983 warship magazine. At MW Books in Ireland, owner Jim Shaughnessy described orders as sporadic, machine-driven and equally eclectic. Another seller told the Guardian that similar buyers had ordered about 6,000 books since January, paying full prices and sometimes routing different aliases to the same Heathrow-area freight warehouses.
Unique insight
The important shift is that AI data acquisition is beginning to look less like a software procurement exercise and more like an industrial supply chain with ordinary small businesses as its sensors.
Analysis
The evidence does not prove that every buyer is working for an AI lab, but the pattern fits a market already established by Anthropic, which has acknowledged acquiring books as part of a broader training-data programme. The immediate effect is commercial: sellers face demand they cannot explain, while buyers can pay premiums for material that is difficult to find online.
For model developers, printed books offer older text that is less likely to contain synthetic AI prose and may fill gaps in digitised corpora. For rights holders, the episode is another warning that training-data disputes are moving beyond crawled websites and licensed databases. In the short term, booksellers will watch buyer identities, delivery addresses and order patterns; in the long term, regulators and courts will have to decide whether opaque physical acquisition is a legitimate data market or a workaround for unresolved copyright questions.
Bottom line — If I only remember one thing tomorrow, what should it be?
The AI training-data race is now creating visible demand in the physical book trade, but who is behind it—and whether the purchases are lawful—remains unproven.
- Story type: Important story
- Impact: Medium
- Confidence: Medium (multiple reports)
- Audience: Business
- What changed: UK and Irish secondhand booksellers reported coordinated-looking bulk orders of unrelated books from opaque buyers.
- What to watch: Whether the named buyers or AI companies disclose the purpose and destination of the books.
- Reading time: 2 minutes
