The Order That Looked Like a Scam
On a summer day in Haarlem, Netherlands, antiquarian bookseller Pieter de Vries received an email from a woman named Natalia, identifying herself as being with a company called '2077AI.' She was requesting a 'fairly large order' — attached was a spreadsheet listing more than 3,001 ISBNs, with instructions to match titles, prepare quotes and estimate shipping costs directed to China.
De Vries deleted it. 'I was shocked!' he told Fortune after a Dutch journalist investigating the request contacted him weeks later and revealed its connection to artificial intelligence. Several other Dutch antiquarian booksellers received the same message and, like de Vries, dismissed it as a scam.
The titles — mostly published between 2020 and 2021 by academic publishers including Emerald Publishing, Elsevier, Wiley, Routledge and Oxford University Press — spanned business, education, engineering, public policy and medicine. The email never mentioned artificial intelligence or explained what the books would be used for.
Destructive Scanning: A Paper Trail With a Paper Shredder
The inquiry did not occur in a vacuum. Last summer, court records revealed that Anthropic had purchased millions of physical books, removed their bindings, scanned them and discarded the originals to build a searchable digital library used to train its AI models — a process internal documents called 'Project Panama,' according to The Washington Post. The technique, known as 'destructive scanning,' involves cutting a book's spine so its pages can be fed through high-speed scanners before the physical copy is thrown away.
A federal judge later ruled that using legally purchased books to train AI models constituted fair use under copyright law. The case settled. Separate claims involving Anthropic's downloading of books from the LibGen and PiLiMi online libraries were also resolved through the settlement.
An Anthropic spokesperson told Fortune that Claude models are trained on a mix of publicly available web data, commercially acquired datasets and internally generated data, and that none of its data acquisition programs acquire or destroy rare or antiquarian books.
An Industry Emerges Around Scale
Earlier this year, 404 Media reported that ISBNdb — a company best known for maintaining metadata tied to International Standard Book Numbers — had begun advertising services to help AI labs source bulk printed book purchases ranging from 1,000 to 1 million books. The company's website said the service was tailored to 'LLM training needs' and 'delivered at the scale AI demands.' Those webpages have since been removed. ISBNdb later told Fortune the service was never launched and the webpages reflected an exploratory concept rather than an active business.
Similar reports have emerged in Germany and Switzerland, where local media reported that secondhand booksellers received unusual bulk orders for highly specialized books. Zoom Books, a Canadian company identified in some of those reports, denied the allegations and told Swiss broadcaster SRF the purchases were part of its regular recycling and trading model.
What the Market Is Really Pricing
The numbers come first: 3,001 ISBNs, millions of physical volumes, and a federal fair-use ruling that effectively handed AI companies a legal roadmap to consume the written output of human civilization at industrial scale. Capital rewards clear rules — and right now the rules favor the data acquirers, not the authors, publishers or booksellers who produced the underlying property.
For free-market readers, the tension here is real. Legally purchased goods used in a legal process is hard to argue with on principle. But when an entire industry quietly builds a physical supply chain designed to strip, scan and discard the products of private enterprise — without disclosing the purpose to sellers — the question of informed consent and property rights deserves more than a settlement and a removed webpage.
