Mysterious bulk orders hit UK bookshops as AI firms race to hoard print data

Secondhand bookshops across the UK and Ireland are reporting a wave of oddly assorted bulk orders, fuelling suspicions that AI companies are quietly sweeping up print books for training data. Over the past few months, booksellers say they’ve received requests from buyers in the US, Canada and continental Europe for hundreds of titles at a time, with no coherent theme—mixing obscure translations, out‑of‑print novels and old specialist magazines in a single order. Some seemingly unrelated buyers even ship to the same logistics hubs near major airports, suggesting a coordinated effort to aggregate and process the books at scale.

Many in the trade now suspect these books are having their spines cut off, being scanned into massive text datasets and then pulped. The concern comes on the heels of revelations that Anthropic, the US AI company behind the Claude chatbot, spent tens of millions of dollars acquiring books precisely for this purpose. For AI labs, high‑quality long‑form text is a scarce and strategically vital resource; for publishers and authors, it raises hard questions about consent, compensation and the preservation of cultural heritage—especially when rare or out‑of‑print works are involved. Expect pressure to grow for clearer rules on how print archives can be used in the age of data‑hungry AI models.