Amazon destroys rare books to feed AI training pipelines

The story's real news isn't that Amazon uses books for training (known) but that the scale and method are now observable via logistics tracking. This means the data sourcing playbook for major labs is becoming transparent in ways it wasn't before, and the destruction of rare materials suggests labs are optimizing for volume over curation.

This is largely disconnected from recent activity in the space we've covered. It belongs instead to the emerging category of AI infrastructure transparency stories, where operational practices become visible through supply chain leaks or forensic tracking. As labs scale foundation models, the sourcing decisions they make (what gets bought, what gets destroyed, what gets preserved) reveal their actual cost-benefit calculations. This matters because it shows labs treating rare intellectual resources as consumable inputs rather than irreplaceable assets, a trade-off that will shape what training data looks like for the next generation of models.

If other major labs (OpenAI, Google, Anthropic) issue public statements about their book sourcing practices or data retention policies within the next 60 days, that signals the practice is becoming a reputational liability. If Amazon's book destruction accelerates rather than slows after this reporting, that confirms the company views the PR cost as lower than the operational benefit of the current approach.