AI Firms Buy Books in Bulk for Training, Sparking Controversy

Recently, multiple book suppliers received unusually large orders, suspected to be from AI companies. These orders are believed to be for obtaining high-quality training data, replacing low-quality internet content. The controversy centers on ISBNdb, a book database site that offered services to connect AI firms with book warehouses, but has since removed the pages and denied involvement. In January, court documents revealed Anthropic's 'Project Panama,' aiming to buy and scan books then destroy them, with a $1.5 billion settlement. Insiders say destroying books is for cost-cutting, not hiding traces. As AI firms need cleaner data, such practices may become more common.