Investigation Reveals Amazon Allegedly Destroying Rare Books for AI Training Data
A recent investigation by Ars Technica has uncovered what appears to be a troubling practice by Amazon: the destruction of rare books that investigators tracked using a hidden AirTag placed inside a shipment.
The investigation centers on Amazon's fulfillment operations, where books were tracked from warehouse facilities to what appears to be disposal sites. The findings suggest that physical books—including rare and potentially valuable editions—are being destroyed rather than recycled, resold, or donated.
The practice, if confirmed, would highlight ongoing ethical questions about how major technology companies obtain training data for AI systems. Large language models and other AI systems require massive amounts of text data, and publishers, authors, and copyright holders have increasingly raised concerns about the unauthorized use of their materials in AI training datasets.
Amazon, whose logo famously features a T. rex preparing to devour a book, has not publicly detailed its data sourcing practices for AI training. The company has invested heavily in AI development across its various services, including Alexa, AWS cloud offerings, and e-commerce features.
This investigation adds to growing scrutiny of AI companies' data practices. Copyright holders and authors' groups have filed numerous lawsuits alleging that tech companies used their works without permission or compensation to train AI systems. The destruction of physical books for this purpose, if verified, would represent a particularly visible manifestation of these concerns.
The investigation raises questions about:
- Supply chain transparency: How Amazon manages excess or returned inventory of rare books
- Data sourcing ethics: Whether physical destruction of books correlates with digital copying for AI training
- Regulatory gaps: Current copyright law's ability to address AI-era content usage
Amazon has not issued a public statement regarding the specific allegations in this investigation.