--- title: Amazon Is Reportedly Destroying Rare Books to Train Its... description: An investigation by 404 Media reportedly found Amazon buying rare, out-of-print books in bulk, destructively scanning them at a facility called VGT3 inside its... canonical: https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Amazon Is Reportedly Destroying Rare Books to Train Its AI Models | daily.dev og:description: An investigation by 404 Media reportedly found Amazon buying rare, out-of-print books in bulk, destructively scanning them at a facility called VGT3 inside its... og:url: https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z og:image: https://api.daily.dev/og/posts/Ux2GhMP3Z.png og:image:alt: Amazon Is Reportedly Destroying Rare Books to Train Its AI Models og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Amazon Is Reportedly Destroying Rare Books to Train Its AI Models **[Collections](https://daily.dev/sources/collections)** · 1 min read · 1 upvotes · 0 comments ## Summary An investigation by 404 Media reportedly found Amazon buying rare, out-of-print books in bulk, destructively scanning them at a facility called VGT3 inside its LAS8 Las Vegas warehouse, then discarding the physical copies. Reporters tracked a shipment of about 1,000 books using a hidden AirTag to confirm the destination. The scanned text is believed to feed AI model training, since out-of-print books offer fresh, unrecycled language that predates the AI era and carries less contamination risk than scraped internet text. Amazon's official response only confirmed it acquires books through normal commercial channels without addressing the destructive scanning claim. ## Content A 404 Media investigation has traced a mysterious bulk purchase of rare and out-of-print books to an Amazon facility in Las Vegas, where workers reportedly cut the spines off books and scan the pages to train AI models. The story started with a bookseller's hunch. When a buyer scooped up around 1,000 books in one order, the seller tucked an AirTag inside one of them before shipping it out. That tracker led 404 Media's reporters to VGT3, a scanning operation inside Amazon's LAS8 warehouse. Posts on Amazon worker forums apparently confirm what the tracking data suggested: the facility destructively scans books at scale, slicing off the binding so pages can be fed through a scanner faster. I find this whole chain of events kind of wild - a physical tracking device solving what amounts to a corporate mystery about AI training data. It's the kind of investigative work that feels almost old-fashioned given the subject matter. Asked about it, Amazon didn't confirm or deny the AI connection. The company said only that it buys books through commercial channels to ## Questions this post answers ### Why would an AI company want to scan rare out-of-print books instead of using internet text? Out-of-print books provide language that has not been recycled across countless existing training sets, since the open internet has largely already been scraped. They also predate the AI era, meaning the text was written by humans without any influence from AI-generated content, which reduces the risk of models training on AI-contaminated data. _Follow how companies source AI training data as debates over data provenance shape AI development on daily.dev._ ### How did reporters confirm Amazon was destructively scanning rare books for AI training? Investigators hid an AirTag inside a shipment of roughly 1,000 rare books being sold in bulk and tracked it to a facility called VGT3 inside Amazon's LAS8 warehouse in Las Vegas. Worker forum posts reportedly described VGT3 as a site where books are destructively scanned at scale, meaning the physical book is destroyed once the text is captured. _Track investigative reporting on AI data sourcing practices for developers weighing AI ethics on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 33 comments across lobsters (as of 2026-08-18). **TL;DR:** Discussion centers on the legality and ethics of destructively scanning purchased books for AI training, with debate over whether the practice is a legitimate (if unsettling) side-effect of copyright law or a tragic waste of physical media. **Sentiment:** 5% positive · 35% mixed · 60% skeptical **The case for** - Some argue the books would have decayed or been landfilled anyway, so scanning gives them a better end than rotting away. - A few note this is simply standard commerce—if a company pays for goods it can use them however it likes. - One commenter points out companies are legally obligated to destroy originals after digitizing, so this isn't necessarily malicious but a copyright-law quirk. **The pushback** - Many find it disturbing that unique/rare content is destroyed via a 'one-way' process that makes the original text irrecoverable to the public. - Skepticism that companies actually delete their scans after a single training run, suspecting they keep copies for future training. - Concern that authors and publishers get nothing since these are secondhand purchases, yet multiple companies must redundantly buy and destroy copies since scans can't be shared. - Some see the whole practice as ethically corrosive even if technically legal, with authorities unlikely to intervene. **By community** - lobsters (mixed): Debate ranges from concern about destructive scanning and copyright absurdities to pragmatic arguments that decaying books are better digitized than left to rot, with no clear consensus. **Hottest debate:** Whether destroying scanned books is a real ethical/legal problem or just misplaced sentimentality given that many of these books would have decayed anyway. **Open questions** - Is planting a tracking device in a shipped product itself legal? - Do AI companies actually delete scanned book files after training, or do they retain them for future use? - Could copyright law be reformed to let digitized-then-destroyed works be shared publicly? **Highlights** > I think it's pretty absurd that we have found ourselves in a situation where an effective and apparently legal way to train AI is to buy *second hand* books - so the author/publisher gets nothing - and truck them around the country, then cut them up and scan them in... and multiple companies need to do the same thing because the scanned work cannot be shared. Is it "worth it"? I have no idea. If the books were going to eventually decompose in a warehouse somewhere (or go into landfill) then I guess this is a better ending for them, but that's a big "if". > — [simonw on lobsters · 2 points](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at#c_axwvwn) > When people hear the word "rare" they think of valuable, collectible, antique, one-of-a-kind, historically-significant, leather-bound books. In these cases it really just means "obscure", and almost always "unwanted". There are literal *tons* of books being weeded from libraries and sold by the pallet. 1964 tractor manual, COBOL for Dummies, Windows NT 4.1 Network Administration... Also keep in mind that mass-market printings only last 40 years or so before deteriorating. Digitizing is actually better for their longevity, rather than rotting into nothing! They certainly do not throw away the scan after training. The real tragedy is copyright law: These companies are 1) legally obligated to destroy the original, 2) legally prevented from sharing the scans. I wonder if AI companies could recover some goodwill by writing a blog post ~"we would love to host a digital library, but are legally prevented from doing so". > — [chadski on lobsters · 1 points](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at#c_x5xcmk) > I am 100% convinced that if you buy some items you are not breaking any laws by disposing of them as you see fit. As long as the book pulping process doesn't release any harmful chemicals, the authorities have no cause to investigate. This entire process is arguably unethical and can cause long-term harms to society, but it is not illegal. > — [gerikson on lobsters · 3 points, 1 comments](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at#c_359rmk) > That's the worst part. These books are digitised and destroyed. The contents are transformed with a one-way algorithm and irrecoverable. What the fuck is wrong with these people? > — [WilhelmVonWeiner on lobsters · 1 points, 1 comments](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at#c_7peva4) > Something that I see a lot of people missing in this discussion (globally, not necessarily here) is that if you want to digitize a work (and use it as otherwise allowed in copyright law) you must destroy the original (https://cdn.arstechnica.net/wp-content/uploads/2025/06/Bartz-v-Anthropic-Order-on-Fair-Use-6-23-25.pdf). > — [benji on lobsters · 1 points](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at#c_ik9p64) **Source threads** - [lobsters](https://lobste.rs/s/flcpeu/we_tracked_shipment_rare_books_it_ended_at) · 51 points · 33 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#amazon](https://daily.dev/tags/amazon), [#ethical-ai](https://daily.dev/tags/ethical-ai) [View this post on daily.dev](https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Amazon Is Reportedly Destroying Rare Books to Train Its AI Models","url":"https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z"},"datePublished":"2026-08-18T10:12:29.231Z","dateModified":"2026-08-18T23:42:59.224Z","description":"An investigation by 404 Media reportedly found Amazon buying rare, out-of-print books in bulk, destructively scanning them at a facility called VGT3 inside its...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cd7c74818ccba4b0a59db3ce7fbf9c0a?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cd7c74818ccba4b0a59db3ce7fbf9c0a?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,amazon,ethical-ai","timeRequired":"PT1M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Amazon Is Reportedly Destroying Rare Books to Train Its AI Models"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/amazon-is-reportedly-destroying-rare-books-to-train-its-ai-models-ux2ghmp3z#faq","mainEntity":[{"@type":"Question","name":"Why would an AI company want to scan rare out-of-print books instead of using internet text?","acceptedAnswer":{"@type":"Answer","text":"Out-of-print books provide language that has not been recycled across countless existing training sets, since the open internet has largely already been scraped. They also predate the AI era, meaning the text was written by humans without any influence from AI-generated content, which reduces the risk of models training on AI-contaminated data. Follow how companies source AI training data as debates over data provenance shape AI development on daily.dev."}},{"@type":"Question","name":"How did reporters confirm Amazon was destructively scanning rare books for AI training?","acceptedAnswer":{"@type":"Answer","text":"Investigators hid an AirTag inside a shipment of roughly 1,000 rare books being sold in bulk and tracked it to a facility called VGT3 inside Amazon's LAS8 warehouse in Las Vegas. Worker forum posts reportedly described VGT3 as a site where books are destructively scanned at scale, meaning the physical book is destroyed once the text is captured. Track investigative reporting on AI data sourcing practices for developers weighing AI ethics on daily.dev."}}]} ```