As the internet becomes inundated with AI-generated content, the availability of 'clean corpora'—datasets created solely by humans—has become increasingly scarce. Against this backdrop, AI giants have begun to set their sights on printed books as a valuable source of untainted data. According to a report by independent media outlet 404 MediaX, AI companies are now purchasing rare books published prior to 2022 in large quantities. They are extracting clean data, free from AI-generated contamination, by physically cutting off the spines of these books and scanning their pages.
