Anthropic purchased millions of physical books to train its AI chatbot, Claude. The arrangement and scope of those acquisitions are a central focus of the reporting. The material described as “destructively scanned” is part of the context surrounding the training process, though the precise outcomes of those scans are not detailed in the provided content. The narrative centers on the books in question and the claimed method described as destructive scanning.
The report frames the issue around a substantial volume of physical books obtained by Anthropic, intended to support Claude’s development. It notes that the process included scans described as destructive, and it labels that aspect as a focal point of the discussion. The facts presented are limited to the existence of these book purchases and the described scanning approach, without further specifics on quantities, titles, or the exact nature of the destruction.
No additional details, quotes, or corroborating information are provided in the source excerpt beyond the characterization of the books and the described scanning method. The content hints at broader conversations about training data and copyright implications but stops short of presenting those arguments or any conclusions.
In short, the material establishes that Anthropic bought millions of physical books for Claude’s training and that some scans were described as destructive. The document does not supply exact figures, titles, or outcomes of the scanning, leaving those elements unverified within the provided RSS content.
