Shafaqna English- Artificial intelligence companies are increasingly purchasing large volumes of physical books, digitizing their contents for AI training and then destroying the originals, raising concerns over the loss of rare and out-of-print works, according to Futurism.
According to reports, several AI developers have shifted from relying solely on digital books to acquiring physical copies, which are scanned using industrial equipment before being dismantled or destroyed. In a recent copyright case involving Anthropic, court documents revealed the company removed book pages with hydraulic cutting machines to create digital copies for training its language models.
The practice relies on the legal “first-sale doctrine,” which allows owners to modify or destroy purchased physical books. A U.S. court ruled that converting printed books into searchable digital files for AI training constituted a transformative use protected under fair-use principles, provided the books were not redistributed as new copies.
The growing demand has created a new market for book suppliers. ISBNdb, a company that previously focused on helping libraries and booksellers manage inventories, now facilitates bulk purchases ranging from thousands to one million books for AI developers while keeping buyers anonymous.
Booksellers say demand has surged, particularly for older and foreign-language titles. Some have expressed concern that rare or out-of-print books—sometimes among the last remaining copies—could be permanently lost after being digitized. While the business has boosted sales and reduced unsold inventory, critics warn the trend could threaten valuable printed works and cultural heritage if it continues on a large scale.
Source: Futurism