All NewsEducationTV
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

Secondhand book sales surge amid AI training concerns

Created at 15 Aug · 11:41 AM1 source↑ Market-relevant
IN SHORT

Independent booksellers worldwide are reporting unusual bulk orders of secondhand books, with suspicions mounting that the purchases are linked to AI companies destructively scanning texts to train large language models. A US court ruling found such use transformative and not a copyright violation, though ethical concerns about book destruction persist.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

30 yearssecond-hand book trade experience of one seller
100 copieslimited print run of a recent academic text
75copies of the academic text already in libraries
18th centurycentury of a rare book sold

Who's Involved

Stuart Manley
Owner of Barter Books, reporting unusual bulk orders
David Tobin
Owner of Walden Books, also experiencing unusual sales
Judge William Alsup
US judge who ruled on AI training and copyright law
Anthropic
AI firm involved in a lawsuit over book usage for training
Claude
AI chatbot developed by Anthropic
Professor Emily Hudson
Intellectual property specialist at Oxford University
Derek Walker
Owner of Edinburgh bookshop McNaughtan's, discussing ethical concerns

↳ Why This Matters

The surge in secondhand book sales, potentially driven by AI companies' demand for training data, highlights a new and ethically complex intersection between literature, technology, and copyright law. It raises questions about the future value and preservation of physical books in the age of AI and the economic impact on independent booksellers.

Key facts

  • Secondhand booksellers are experiencing a surge in bulk orders from unknown buyers.
  • The purchases are suspected to be for AI companies to use books as training data.
  • A US court ruling in 2025 deemed the use of purchased books for AI training as 'exceedingly transformative' and legal.
  • Internal documents from an AI firm referred to a project aiming to 'destructively scan all the books in the world'.
  • Destructive scanning involves digitizing books by removing pages, often leading to the book's destruction.
  • Booksellers are concerned about the ethical implications of books being destroyed, particularly rare editions.
  • UK copyright law requires permission from copyright owners for creating training libraries.

A significant increase in secondhand book sales, characterized by large, unusual bulk orders, has been observed by booksellers globally. This phenomenon has led to speculation that the demand is being driven by artificial intelligence companies seeking to acquire vast quantities of texts for training their large language models. Booksellers like Stuart Manley of Barter Books have reported receiving orders equivalent to a week's sales in a single transaction, a trend unprecedented in their decades of experience.

The suspicion that AI firms are behind these purchases stems from a US court ruling in 2025. In a lawsuit brought by three authors against AI firm Anthropic, a judge determined that using purchased books to train AI software was permissible under US copyright law, deeming the use 'exceedingly transformative.' Unsealed court documents further revealed that the process of ingesting books for training Anthropic's AI chatbot, Claude, involved 'destructive scanning,' where books are digitized at an industrial scale, often by removing their spines before being recycled.

Anthropic stated that sourcing books is a common industry practice and that their data acquisition programs do not target rare or antiquarian books. However, the revelation of destructive scanning has caused unease among booksellers. David Tobin of Walden Books expressed concern that books, even those not rare, might ultimately be destroyed. Experts suggest that the diverse and sometimes obscure nature of the books being purchased could be an effort to provide fresh material to improve the training of large language models.

Copyright laws in the UK differ from those in the US, with the starting point being that copying acts for training libraries require the copyright owner's permission, according to Professor Emily Hudson, an intellectual property specialist at Oxford University. This presents an ethical dilemma for booksellers, who are uncomfortable with the idea of books being destroyed, especially unique or historically significant editions. While some books may not be considered essential to preserve, the potential destruction of a sole surviving copy of an 18th-century edition, for instance, is a significant concern. Some booksellers, however, acknowledge that recycling unwanted books can be beneficial, especially when it boosts trade, with one noting that 'the world no longer needs five million copies of The Da Vinci Code.'

Frequently asked questions

Independent booksellers are observing a significant rise in bulk orders, leading to speculation that AI companies are purchasing books to use as training data for their models.

Destructive scanning is a process where books are digitized at an industrial scale, often involving removing the book's spine to rapidly scan pages, after which the book is typically recycled.

A US court ruled in 2025 that using purchased books for AI training was 'exceedingly transformative' and thus not a violation of US copyright law. However, UK copyright law may differ.

While AI firms state they do not acquire rare or antiquarian books for destruction, the general practice of destructive scanning raises concerns among booksellers about the potential loss of unique or historically significant texts.

What Happens Next

01Further legal challenges or legislative action regarding AI training data acquisition may arise.
02AI companies may continue to source books, impacting the secondhand market and potentially leading to further ethical debates.
03Booksellers may face ongoing dilemmas regarding sales to entities involved in destructive scanning.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

Independent booksellers are observing a pattern of unusually large bulk orders for secondhand books.
Booksellers suspect these bulk orders are for AI companies to train large language models.
A US court ruled in 2025 that using purchased books to train AI software was not a copyright violation.
Internal company documents revealed an AI firm's aim to 'destructively scan all the books in the world'.
The process involves digitizing books at scale, often by removing spines and recycling the remains.
Booksellers express unease about the potential destruction of books, especially rare or unique editions.
Experts suggest the diverse subject matter of books being purchased could be to improve AI training data.
UK copyright laws differ, generally requiring permission from copyright owners for such copying.

Sources

T1
Secondhand book sales are booming. Is it because of AI?BBC News

Related Stories

OpenAI, Anthropic Cut Prices as Chinese AI Rivals Gain Ground
14 Aug · 2:31 PM
Man attempts to manipulate court with hidden AI prompts in filings
14 Aug · 5:31 PM
Wall Street Banks Invest Billions in AI for Trading, Operations, and Culture
15 Aug · 10:36 AM
LinkedIn AI cringe slop generator Cringebot3000 gains traction
15 Aug · 10:11 AM
China's EVs Get Smarter With AI-Powered Features
15 Aug · 8:06 AM