[CLSTR]
< Back to situations

Monitor this situation.

[SITUATION] · [QUIET] · [TECHNOLOGY]

10 clusters · 94 sources · 78 days · First seen · Last updated

AI companies accused of destructive book scanning

Overview

Concerns have emerged regarding the practice of ‘destructive scanning’ used to acquire training data for artificial intelligence. Initially, Australian booksellers and rare-book experts reported that the Canadian company Zoom Books was making large, irregular orders for niche, decades-old titles, expressing concern that these books were being destroyed after digitization. Scrutiny has since expanded to the AI developer Anthropic. Court documents regarding an internal initiative called ‘Project Panama’ suggest the company has been acquiring millions of secondhand and out-of-print books to facilitate digital data collection. Reports indicate that Anthropic uses a ‘scan-and-destroy’ method to train its Claude AI, where book spines are slit and pages are trimmed to meet scanning formats, rendering them unusable. An internal memo allegedly describes the objective as ‘destructively scanning all the books in the world.’ Investigations have also identified a specialized system, ‘Nippan Loader,’ used to acquire and digitize Japanese books. Anthropic reportedly contracted with Nippan, Japan’s largest book wholesaler, to purchase large quantities of physical books for shredding and scanning. This has coincided with a surge in orders for niche professional texts in Japan, with trade data suggesting over 50 tons of Japanese books have been exported to the United States. The Japan Book Publishers Association has formally requested an explanation from Nippan, as the situation sparks debate over whether such practices constitute ‘unreasonable harm’ to copyright holders. Investigations by 404 Media have expanded the scope of these allegations to Amazon. Using an Apple AirTag to track a shipment of 1,000 books, investigators located an Amazon warehouse in Las Vegas (VGT3) where workers reportedly received large shipments to strip ISBN barcodes and covers before destruction.

Entities

Anthropic · Project Panama · 404 Media · Amazon · Biblio

Claims

What the coverage asserts, and how many sources carry each claim.

Coverage disagrees

Sources make claims that cannot both be true. CLSTR reports the disagreement; it does not decide who is right.

  • "Anthropic claims that no rare books are destroyed through this method."

    vs

    "The process involves cutting the spines off books to speed up digitization, which destroys the original copies."

    Anthropic claims no rare books are destroyed, whereas industry analysis asserts the process destroys original copies by cutting spines.

Timeline

  1. [TECHNOLOGY] 2 sources
    Anthropic to pay $1.5 billion to settle AI copyright lawsuit

    Anthropic will pay $1.5 billion to settle a US copyright lawsuit involving 500,000 books used for AI training, including works by prominent South Korean authors like Han Kang.

  2. [TECHNOLOGY] 2 sources
    Anthropic settles copyright lawsuit for $1.5 billion

    Anthropic has reached a $1.5 billion settlement in a legal battle over the use of copyrighted books to train its AI models, following a court ruling that categorized such use as fair use.

  3. [TECHNOLOGY] 4 sources
    Anthropic used specialized system to scan Japanese books for AI training

    Court filings and trade data reveal Anthropic used a specialized system to acquire and scan massive quantities of Japanese books for AI training, sparking copyright concerns in Japan.

  4. [TECHNOLOGY] 3 sources
    Anthropic linked to physical book destruction for AI training

    Anthropic’s ‘Project Panama’ has reportedly involved purchasing and destroying millions of physical books to scan their content for AI training, sparking concerns over cultural memory loss.

  5. [TECHNOLOGY] 3 sources
    Amazon scans and destroys books for AI training in Las Vegas

    An Amazon warehouse in Las Vegas is reportedly scanning and destroying books, including rare and library copies, to collect data for training artificial intelligence systems.

  6. [TECHNOLOGY] 10 sources
    Anthropic faces scrutiny over ‘Project Panama’ book destruction

    A coalition of US civil rights groups is calling for an FTC investigation into AI company Anthropic’s ‘Project Panama’, which involves the destructive scanning and disposal of physical books to train AI models.

  7. [TECHNOLOGY] 8 sources
    AI companies accused of buying and destroying books for training data

    AI companies are reportedly buying and destroying vast quantities of physical books, including rare editions, to scan them for training large language models.

  8. [TECHNOLOGY] 33 sources
    Amazon uses destructive scanning of rare books to train AI

    An investigation revealed Amazon is buying rare books and using destructive scanning at a Las Vegas facility to train its AI models, specifically targeting pre-2022 texts to ensure high-quality data.

  9. [TECHNOLOGY] 10 sources
    Anthropic accused of destroying books for AI training

    AI firm Anthropic is accused of using ‘destructive scanning’ to train models, involving the mass purchase and physical destruction of secondhand books to facilitate rapid digitization.

  10. [TECHNOLOGY] 3 sources
    Australian Booksellers Warn AI Over Zoom Books' Rare Book Purchases

    Australian booksellers warn that Canadian firm Zoom Books is buying and possibly destroying rare books to feed AI training data, sparking concerns over cultural loss.

Sources

agora-web.jp · aktuelno.me · anzca.edu.au · apple.slashdot.org · autogpt.net · bigdata-insider.de · bild.de · bilder3.n-tv.de · bildungsserver.de · bitcoinethereumnews.com · bizlife.rs · biznismagazin.rs · borncity.com · cafef.vn · creativebloq.com · cryptobriefing.com · diariodejerez.es · diariodesevilla.es · digital.t-online.de · dnevno.hr · dnyuz.com · doktus.de · doviz.com · editionblaes.de · eldiadecordoba.es · eldiariony.com · espreso.co.rs · europasur.es · extratipp.com · fehmarn24.de · finance.technews.tw · gamemag.it · gamestar.de · geekwire.com · generation-nt.com · giga.de · glas-slavonije.hr · greekcitytimes.com · hlavnespravy.sk · hna.de · index.hr · inmedia.ba · internet.watch.impress.co.jp · iphonesoft.fr · it-boltwise.de · itbiz.cz · ithome.com · itnetwork.rs

This summary has been updated 13 times: see revision history