LIVE
+++ Anthropic revenue run rate surges past $65 billion +++ OpenAI plans 8 GW Ohio data center – Nvidia backs $105B +++ Groq raises $350M for its neocloud pivot +++ AirTag probe: Amazon destroys rare books for AI training +++ AI startup Relay shuts down – team joins Google Chrome +++ Penn State study: only 17% of user rules survive compaction ++++++ Anthropic revenue run rate surges past $65 billion +++ OpenAI plans 8 GW Ohio data center – Nvidia backs $105B +++ Groq raises $350M for its neocloud pivot +++ AirTag probe: Amazon destroys rare books for AI training +++ AI startup Relay shuts down – team joins Google Chrome +++ Penn State study: only 17% of user rules survive compaction +++
All news ›
AI IN LIFE AI IN LIFENEWS
DAILY
SOCIETY

AirTag reveals Amazon is destroying rare books to train AI

A 404 Media investigation shows Amazon buying rare books, slicing off their spines and scanning them as training material for AI models.

AirTag reveals Amazon is destroying rare books to train AI

Illustration · AI-generated (AI IN LIFE)

At a glance

  • 404 Media tracked a tagged book via AirTag to Amazon's VGT3 facility in Las Vegas
  • Book spines are cut off; pages are scanned destructively
  • Amazon: purchases are made "through commercial channels" to improve products
  • Pre-2022 texts are considered free of AI output – key against model collapse
  • Context: Anthropic paid a $1.5B settlement in 2026 over pirated books

Amazon – the company that started out selling books online – is reportedly buying up rare and out-of-print books, cutting them apart and scanning them to train AI models. The practice was uncovered by 404 Media: journalists hid an Apple AirTag inside a rare book and tracked it to an Amazon facility in Las Vegas known as VGT3.

According to TechCrunch, the facility is marked by a dinosaur logo holding a book. There, book spines are cut off at scale and pages are machine-scanned – a destructive process that leaves the books unusable.

Amazon told 404 Media only that it "purchases books through commercial channels to improve the products and services customers use." The company declined to comment on volumes, titles, or how it handles particularly rare copies.

The appetite for printed originals has a technical reason: freely available internet text is already in the models – and texts from before 2022 are considered "clean" because they are guaranteed not to be AI-generated. That lowers the risk of so-called model collapse, where models degrade by ingesting AI output.

The practice echoes the Anthropic case: that company scanned millions of books and, over previously pirated copies, agreed to a $1.5 billion copyright settlement in July 2026. Amazon's purchases reportedly run through commercial channels – but the debate over the cultural loss of rare works is likely to intensify.

◈ AI-GENERATED REPORT · SOURCES LINKED

FAQ

How was the practice uncovered?

404 Media hid an Apple AirTag in a rare book and tracked it to an Amazon scanning facility in Las Vegas (VGT3).

Why does Amazon destroy the books?

Destructive scanning is faster and cheaper. Rare, older texts are valuable training material because they are guaranteed not to be AI-generated.

Is this legal?

Amazon says it buys the books through commercial channels – unlike the Anthropic case, where pirated copies led to a $1.5B settlement. Copyright questions remain open.