Amazon's VGT3 Facility Found Destroying Rare Books for AI Training Data

The destruction of rare books for AI training raises ethical concerns and highlights the competitive pressures in the tech industry.

• On August 17, 2026, reports surfaced that Amazon's VGT3 facility in Las Vegas is destroying rare books for AI training.
• An Airtag embedded by a bookseller tracked a shipment of rare books to the facility, confirming suspicions of systematic de-spining and scanning.
• Amazon's response was vague, focusing on book purchases for product improvement without addressing the AI training specifics.

• AI developers require vast amounts of unique textual data to train advanced models, leading to increased competition for original content.
• Booksellers have noted bulk purchases of rare and out-of-print works by unidentified buyers since at least 2025, indicating a growing trend.
• Rivals like Anthropic and xAI have avoided antique books, creating pressure on Amazon to secure unique data sources not available in public datasets.

The revelation that Amazon is destroying rare books for AI training has sparked significant debate about the ethical implications of such practices. The need for unique, non-public domain textual data has become increasingly critical in the competitive landscape of AI development. Companies like Amazon are under pressure to deliver cutting-edge AI models that can outperform rivals such as OpenAI and Anthropic. This competitive environment has led to a strategic pivot towards acquiring original content, including rare and out-of-print books.

Booksellers have long suspected that their inventory was being targeted for destruction, with reports of bulk purchases by unidentified buyers surfacing as early as 2025. The tracking operation conducted by 404 Media confirmed these suspicions, revealing that rare books were being sent to Amazon's VGT3 facility, where they were systematically de-spined and scanned for text extraction. This process not only destroys the physical copies but also eliminates the historical and sentimental value associated with these works.

Amazon's vague response to the allegations—focusing on the purchase of books for product improvement—does little to assuage concerns. The company has not provided specific details about the number of books being processed or the criteria for selection, leaving many questions unanswered. The lack of transparency has fueled online discussions about the ethical implications of using rare books as training data, with many arguing that the loss of these works represents a significant cultural and historical loss.

As AI continues to evolve, the demand for unique data sources will likely increase, prompting other companies to consider similar practices. This trend raises important questions about the future of literary preservation and the value placed on physical books in an increasingly digital world. The implications extend beyond the realm of AI development, touching on issues of intellectual property, cultural heritage, and the responsibilities of tech companies in safeguarding historical artifacts.

• Booksellers: Concerned about the loss of inventory and historical value.
• Publishers: Facing potential shortages of unique content for future publications.
• AI Developers: Competing for access to unique datasets to enhance model performance.
• Cultural Institutions: Worried about the preservation of literary heritage.

• Regulatory responses: Monitor for any governmental actions regarding the destruction of cultural artifacts for commercial purposes.
• Public sentiment: Watch for shifts in consumer attitudes towards companies that engage in similar practices, which could impact brand loyalty.
• Market trends: Keep an eye on how other tech companies approach data acquisition and whether they follow Amazon's lead or adopt more ethical practices.