Pieter de Vries, a Dutch antiquarian bookseller, initially dismissed an email from '2077AI' in 2026. The request for over 3,000 specific ISBNs seemed like spam, yet it signaled a global AI training operation. This incident reveals a new challenge for physical bookstores and raises concerns about the authenticity of AI-generated content.
The demand for physical books from unexpected sources is rising, but this demand often leads to the destruction of those very books for digital extraction.
Based on covert acquisition tactics and destructive scanning, AI companies will likely continue to aggressively consume physical literary works, potentially eroding the value and authenticity of human-created content long-term.
A Spreading Pattern of Suspicious Orders
Pieter de Vries's experience was not isolated. Other Dutch antiquarian booksellers received identical bulk requests from '2077AI' and dismissed them as well, Fortune reports. The coordinated outreach of identical bulk requests from '2077AI' to other Dutch antiquarian booksellers extends globally. Australian booksellers have reported unusual orders from Zoom Books, a Canadian "book recycler," through platforms like Abebooks or Biblio, according to The Guardian. The simultaneous appearance of these peculiar demands across continents reveals a systematic, international strategy to acquire physical texts. Companies like '2077AI' and 'Zoom Books' exploit the fragmented global secondhand market, using price insensitivity and vague corporate identities to build a covert supply chain for AI training data. Booksellers become unwitting participants in the destruction of cultural artifacts. This widespread pattern implies a deliberate effort to bypass traditional acquisition channels and obscure the ultimate fate of these books.
The Unconventional Nature of AI's Book Hunt
'2077AI' requested 3,001 specific titles, mostly academic works published between 2020 and 2021, Fortune states. The focus on recent scholarship, with '2077AI' requesting 3,001 specific titles mostly academic works published between 2020 and 2021, indicates a need for specialized, current knowledge. In Australia, Zoom Books' orders were price-insensitive, including random, niche, and decades-old stock, The Guardian reports. Zoom Books' price-insensitive orders, including random, niche, and decades-old stock, go beyond typical customer purchases or resale. The peculiar selection and disregard for market price strongly imply these books are acquired for their textual content, not their physical value. The systematic acquisition of both recent academic texts and decades-old niche titles reveals AI's insatiable demand for a comprehensive, multi-temporal knowledge base. The systematic acquisition of both recent academic texts and decades-old niche titles suggests no published work, regardless of age or popularity, is safe from acquisition and destruction for training purposes. The implication is a future where the physical archive of human knowledge is systematically dismantled to feed algorithms.
