A week after Anthropic’s $1.5 billion copyright settlement was approved, new details about how the company sourced its training data have reignited public backlash. Court records and subseque
A week after Anthropic’s $1.5 billion copyright settlement was approved, new details about how the company sourced its training data have reignited public backlash. Court records and subsequent reporting revealed that Anthropic bought millions of physical books in bulk, sliced off their spines using industrial cutting machines, and scanned every page before shredding the originals — a program that ran under the internal name Project Panama.
The practice itself was legally cleared: a federal judge ruled in the underlying case, Bartz v. Anthropic, that digitizing legally purchased physical books and destroying the originals constitutes transformative fair use, since the process didn’t create additional unauthorized copies beyond the one already owned. Separate claims involving Anthropic’s use of pirated books downloaded from LibGen and PiLiMi were resolved through the $1.5 billion settlement, which requires Anthropic to pay roughly $3,000 per book to thousands of authors, according to Fortune’s reporting — the largest copyright settlement in U.S. history.
The $1.5 billion figure made headlines when it was announced, but the visceral image of a hydraulic cutting machine slicing the spine off a purchased book proved to be the detail that actually spread. Investor Michael Burry posted “Evil incarnate” on X in response, and Elon Musk said he’d asked his own SpaceXAI team to preserve rare books in a library and scan them non-destructively rather than cutting off the spine. A Dutch bookseller told Fortune he initially assumed a request for 3,000 copies of the same title was spam or a phishing attempt — a small but telling sign of how unusual bulk book-buying for AI training has become for people on the other side of these transactions.
An Anthropic spokesperson told Fortune that Claude models are trained on a mix of publicly available web data, commercially acquired datasets, and internally generated data, and that the company purchases books through regular commercial markets — explicitly denying that any of its data acquisition programs target or destroy rare or antiquarian books specifically. That statement addresses sourcing intent but doesn’t dispute the underlying mechanics: court records confirm the destructive scanning method was used, and that Anthropic engaged outside vendors, including Datamation Information Services, to perform the high-speed, non-reversible scanning at scale.
Anthropic is not alone in this practice — reporting indicates other AI labs have used similar destructive-scanning services, and brokers now sell comparable bulk book-buying arrangements to AI firms more broadly, sometimes through intermediary platforms like ISBNdb that keep the ultimate buyer anonymous. Pre-2022 books carry particular value in this market, since they predate the recent flood of AI-generated text and offer cleaner, verifiably human-written training data — an economic logic that likely means this sourcing pattern continues industry-wide even as public sentiment toward it sours.
Disclaimer: This content is meant to inform and should not be considered financial advice. The views expressed in this article may include the author’s personal opinions and do not represent Times Tabloid’s opinion. Readers are advised to conduct thorough research before making any investment decisions. Any action taken by the reader is strictly at their own risk. Times Tabloid is not responsible for any financial losses.
The post Anthropic’s Book-Shredding Program Is Going Viral Again — Here’s What Actually Happened appeared first on Times Tabloid.
