WorkAugust 17, 2026via Simon Willison
We Tracked a Shipment of Rare Books. It Ended at an Amazon AI Training Facility
Why it matters
This investigative finding exposes a blind spot in how frontier labs source training data: the systematic acquisition of copyrighted materials (rare books) without transparent licensing or author consent. It matters to practitioners building datasets and to enthusiasts tracking the infrastructure behind model training—and it signals a potential regulatory flashpoint.
Key signals
- Rare book shipment traced to Amazon AI training facility
- Raises copyright and data-sourcing ethics concerns
- Reflects ongoing tension between model makers and content creators over training-data acquisition
- Published by Simon Willison (trusted AI researcher/journalist)
- Investigative tracking methodology suggests systematic procurement pattern
The hook
Amazon is quietly acquiring rare books for AI training—raising new questions about data sourcing ethics and copyright.