Observed Signal · Aug 17, 2026 · Journalistic Investigation · Source: techcrunch · Impact: 3/5 · Sentiment: Neutral

Amazon destroys rare books to train AI models

Executive Signal Summary

TechCrunch reports that Amazon is buying rare, out-of-print books, removing their spines and scanning them for AI training at a Las Vegas facility known as VGT3, based on an investigation by 404 Media that tracked a shipment using a hidden device. Amazon told 404 Media it purchases books through commercial channels to improve customer-facing products and services. The piece notes rare books are valuable training data because they are not available online and predate AI-generated text, and references broader industry issues around data sourcing and copyright, including a recent Anthropic copyright settlement.

Polaris7 AgentPolaris7 Strategic Assessment
High Confidence

Reveals how a major retailer (Amazon) is sourcing rare, offline texts for LLM training, raising data-governance, copyright, and provenance concerns relevant to AI model quality and industry ethics.

SIGNAL RADAR

Track Amazon Signals & Market Shifts in Real-Time

Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.

Start Free in Explorer
Free Explorer tierNo credit card requiredInstant watchlist setup

Key Takeaways & Evidence Grounding

  • 404 Media placed a tracking device in a rare book that ultimately arrived at an Amazon facility in Las Vegas.
  • TechCrunch reports Amazon is buying rare books, cutting off their spines, and scanning them for AI training at a facility identified as VGT3.
  • Amazon told 404 Media it “purchases books through commercial channels to improve the products and services customers use.”
  • The article states rare and out-of-print books are valuable training data for LLMs because they are not available online and predate AI-generated text.
  • The article references Anthropic’s recent $1.5B copyright settlement as context for industry data-sourcing and copyright concerns.

Connected Companies & Entities

4 Entities mapped

“Amazon is buying tons of rare books, cutting off their spines, and scanning them for AI training, according to 404 Media, which placed a tra...”

“Amazon is buying tons of rare books, cutting off their spines, and scanning them for AI training, according to 404 Media, which placed a tra...”

“Companies like Amazon need unfathomably large amounts of text to train their LLMs, which have already ingested what they can from the intern...”

Primary Source Grounding & Direct Attribution
Direct Origin Attribution
Primary Reporting: techcrunch•Published: Aug 17, 2026
Original Coverage Title: “Amazon, once an online bookseller, is destroying rare books to train AI models”

Related Market Signals & Shifts

Recent verified developments and strategic activity across this market segment.

Large Language Models & AI data sourcingAug 27, 2026

Amazon scans and destroys books at Vegas AI warehouse

A t3n report, citing an investigation by 404Media and anonymous Amazon employees, says a large Amazon warehouse in Las Vegas called VGT3 is being used to destroy books and scan individual pages to generate AI training data. Employees describe book spines being cut, pages fed through automated high-speed scanners, and loose pages mixed together and discarded. 404Media tracked large bulk purchases of books routed to the facility (including a July 2026 order of 1,000 books via marketplace Biblio). Amazon told 404Media it buys books through commercial channels to develop its products and services.

Read assessment
Large Language Models (LLM) & AIAug 18, 2026

AirTag Tracks Rare Book Order to Amazon AI Facility

A t3n article reports that 404 Media reporter Emanuel Maiberg hid an Apple AirTag in a bulk order of older books and traced the shipment through US logistics to an Amazon facility in Las Vegas. The investigation suggests companies are buying out-of-print, pre-digital books—which are valuable as human-created training data—to feed AI models. Sellers reported large orders (around 1,000 books) placed without price haggling, and the tracked package traveled from California to a Wisconsin distribution center before arriving at the Amazon LAS8/VGT3 site. Amazon told 404 Media it buys books through commercial channels to improve products and services. The piece references a June 2025 court context where similar book-sourcing practices by Anthropic were judged as "fair use," though Anthropic settled a related dispute for $1.5 billion.

Read assessment
Large Language Models (LLM) & AIJun 24, 2026

Reports: Zoom Books Buying and Destroying Old Volumes

European antiquarian booksellers have reported unusual bulk purchases of out-of-print books in recent weeks. Observers and forum posts suggest a Canadian firm called Zoom Books is buying pallets of older non-fiction titles (typically one copy per ISBN), scanning them into searchable PDFs and then destroying the physical originals — a workflow some AI companies have used to legally claim 'fair use' for training models. The article references a settlement in which Anthropic (developer of Claude) agreed in Sept 2025 to pay $1.5 billion to affected authors over alleged unlawful use of copyrighted texts. Reports say Zoom Books may hold up to 700,000 German titles in a Saxony warehouse and around three million books worldwide, while Zoom Books denies using the texts to train AI but refuses to disclose buyers.

Read assessment

Track Real-Time Market Signals & Shifts

Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.