AI companies are shredding rare books (xcancel.com)

AI companies are engaging in a controversial practice of purchasing rare books, scanning them, and then shredding the originals using high-speed machines. A service called ISBNdb is facilitating these bulk orders, allowing users to buy up to a million books anonymously, emphasizing that pre-2022 texts are particularly desirable because they lack AI-generated content. A recent federal court ruling deemed this practice fair use, as it results in only one copy existing at a time. Amidst this trend, Anthropic has hired the former head of Google Books partnerships to source "all the books in the world." This practice raises significant ethical concerns within the AI/ML community, particularly given the historical and cultural value of the texts being destroyed. The destruction of rare books—some of which are the last remaining copies of their kind—illustrates a troubling shift towards prioritizing data acquisition for AI training over the preservation of irreplaceable knowledge. Critics argue that while scraping data from the internet or reprinting popular texts is reversible, the shredding of unique historical documents creates a permanent loss. With the court's backing, this irreversible trend may accelerate, prompting urgent discussions about intellectual property rights and the ethics of data sourcing in AI development.