This situation has concluded
It was preserved as a record on October 5; the timeline below shows how it unfolded, with sources. Get the briefing to follow the top situations still developing: three emails a week, sourced and in order.
Unsubscribe anytime.
[SITUATION] · [QUIET] · [TECHNOLOGY]
8 clusters · 62 sources · 40 days · First seen · Last updated
AI book digitization and destructive scanning controversy
Overview
In late July 2026, SpaceXAI announced it would use non-destructive scanning for rare books, contrasting with the high-speed, spine-cutting methods used by other firms. This follows reports into Anthropic’s ‘Project Panama,’ which involved purchasing large batches of books, cutting their spines for industrial-speed scanning, and destroying the paper copies. While a U.S. federal judge previously ruled that scanning legally purchased books can qualify as fair use, Anthropic later settled related lawsuits for $1.5 billion.
By mid-August 2026, scrutiny intensified as booksellers in Europe, Australia, and North America reported unusual, large-scale orders of eclectic titles. Experts suggest these acquisitions aim to secure high-quality, human-written text to prevent ‘model collapse’—a phenomenon where AI models degrade by training on their own synthetic output. New investigations by 404 Media used an Apple AirTag to track a shipment of books to an Amazon-owned warehouse, where employees reported scanning books as a primary task. Reports also indicate that facilities, such as an Amazon scanning site in Las Vegas, have been observed cutting spines to facilitate rapid digitization.
In early September 2026, the pattern of unusual acquisitions expanded to include specialized academic content. Bookstores in Canada reported surges in bulk orders for older, technical volumes covering subjects such as medicine, mathematics, and physics. These orders often consist of random lists of titles that have historically remained unsold. In several documented cases, books ordered from Toronto-based retailers were shipped to a logistics company in Illinois, further fueling suspicions that AI companies are targeting specialized texts for training data.
Entities
Anthropic · Project Panama · Zoom Books · SpaceXAI · U.S. federal court
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 10 SOURCES] AI companies are buying large numbers of printed books published before 2022 for training data.
- [● 9 SOURCES] The scanning process involves cutting off the book spine, destroying the original copy.
- [● 8 SOURCES] A U.S. federal judge ruled that scanning legally purchased books can be considered fair use, but using pirated copies is not.
- [● 7 SOURCES] AI companies are buying large quantities of used, old books to digitise for training language models.
- [● 7 SOURCES] Anthropic ran an internal initiative called “Project Panama” to acquire and scan millions of books, aiming to scan up to two million books.
- [● 7 SOURCES] ISBNdb supplies physical books in bulk to AI developers for training purposes.
- [● 6 SOURCES] AI companies are buying large quantities of used physical books to train large language models.
- [● 6 SOURCES] Anthropic settled lawsuits over pirated books for $1.5 billion.
- [● 6 SOURCES] Critics warn that rare and out‑of‑print books are being destroyed, causing cultural loss.
- [● 6 SOURCES] AI companies are purchasing large quantities of used, old books to digitise for training language models.
- [● 6 SOURCES] The scanning process involves cutting off the book spine, scanning pages, and destroying the original copies.
- [● 6 SOURCES] Critics warn that the practice threatens rare and out‑of‑print books, causing cultural loss.
- [● 5 SOURCES] A U.S. federal judge ruled that scanning legally purchased books into digital copies, even when the originals are destroyed, constitutes fair use.
- [● 5 SOURCES] The process called “destructive scanning” cuts book spines, scans pages with industrial scanners, and discards the physical copies.
- [● 5 SOURCES] Anthropic’s internal programme “Project Panama” involved purchasing and scanning millions of books.
- [● 5 SOURCES] AI companies are buying pre‑2022 printed books because they are free of AI‑generated text.
- [● 4 SOURCES] The practice is motivated by the need for clean data not influenced by AI‑generated text.
- [● 4 SOURCES] The practice raises copyright concerns and is part of a broader debate on AI data sourcing.
- [● 4 SOURCES] ISBNdb is offering AI companies the procurement of up to one million physical book titles for training data.
- [● 4 SOURCES] AI companies are purchasing large quantities of used, old books to digitise them for training language models.
- [● 4 SOURCES] The process called “destructive scanning” involves cutting book spines, scanning all pages with industrial scanners, and discarding the physical copies.
- [● 4 SOURCES] Anthropic’s internal programme “Project Panama” aims to acquire and digitise millions of printed books using the destructive‑scanning method.
- [● 4 SOURCES] Anthropic has spent tens of millions of dollars to procure millions of books for Project Panama.
- [● 4 SOURCES] Book‑sellers, collectors and cultural‑heritage specialists in Europe and North America have raised concerns about the practice.
- [● 4 SOURCES] Anthropic argues that the scanning constitutes transformative use under U.S. copyright law.
- [● 4 SOURCES] The digitised text from the scanned books is used to train Anthropic’s AI models, including Claude.
- [● 4 SOURCES] Anthropic ran an internal initiative called “Project Panama” to scan up to two million books.
- [● 4 SOURCES] Companies use hydraulic cutting machines to cut book spines before scanning, then recycle the remnants.
- [● 4 SOURCES] AI companies are buying large numbers of physical books published before 2022 for training data.
- [● 3 SOURCES] A U.S. federal judge ruled that scanning legally purchased books and destroying the originals can be considered fair use.
- [● 3 SOURCES] Booksellers in Europe and North America report a surge in bulk orders for rare, out‑of‑print titles from AI firms.
- [● 3 SOURCES] AI companies are purchasing printed books published before 2022 for training data.
- [● 3 SOURCES] The books are scanned by cutting off their spines, after which the original copies are destroyed or recycled.
- [● 3 SOURCES] AI companies are buying large numbers of physical books for AI training.
- [● 3 SOURCES] The books are cut, scanned, and the originals are shredded or recycled.
- [● 2 SOURCES] The practice has raised concerns about loss of rare and out‑of‑print books and cultural preservation.
- [● 2 SOURCES] Dutch bookseller Pieter de Vries received a bulk list of ISBNs from an AI firm, indicating a large‑scale acquisition effort.
- [● 2 SOURCES] Anthropic spent tens of millions of dollars on Project Panama.
- [● 2 SOURCES] ISBNdb offers AI companies procurement of up to one million physical book titles for training data.
- [○ 1 SOURCE] Anthropic previously pirated more than seven million books from LibGen and Books3 before switching to purchases.
- [○ 1 SOURCE] Zoom Books, a Canadian‑American firm, orders large volumes of older books from German antiquarian markets, often 800‑1,200 copies per title and up to 10,000 copies per order.
- [○ 1 SOURCE] OpenAI, Google and Meta are also facing lawsuits over their methods of acquiring training data.
- [○ 1 SOURCE] Anthropic’s Project Panama aimed to scan up to two million books.
- [○ 1 SOURCE] Anthropic ran an internal initiative called “Project Panama” to acquire and scan books.
- [○ 1 SOURCE] Anthropic spent tens of millions of dollars on the book‑acquisition and scanning operation.
- [○ 1 SOURCE] The targeted books are primarily published before 2022 to avoid AI‑generated content.
- [○ 1 SOURCE] A U.S. federal judge ruled that scanning legally purchased books for AI training is fair use.
- [○ 1 SOURCE] The same judge found that use of illegally copied books from shadow libraries is not protected, leading to a $1.5 billion settlement.
- [○ 1 SOURCE] Australian booksellers reported a surge in sales of rare books due to AI labs' purchases but warned about shredding.
Timeline
-
[TECHNOLOGY] 4 sourcesSpecialized textbooks see bulk purchases linked to AI training
Canadian booksellers report unusual bulk orders of specialized technical and academic books, suspected to be for training artificial intelligence models.
-
[TECHNOLOGY] 2 sourcesAI-generated content rises as tech firms digitize physical books
AI-generated content now accounts for roughly 35% of new websites. To prevent model collapse, tech companies are reportedly digitizing and destroying physical books to secure human-authored training data.
-
[TECHNOLOGY] 2 sourcesAI developers face scrutiny over book destruction and web crawling practices
AI developers are facing scrutiny for destroying physical books to scan them for data, while Meta’s web crawlers have emerged as the dominant force in AI-driven web scraping.
-
[TECHNOLOGY] 18 sourcesAI companies suspected of driving secondhand book sales boom
Secondhand booksellers worldwide report a surge in bulk orders suspected to be driven by AI companies using destructive scanning to acquire human-written text for model training.
-
[TECHNOLOGY] 3 sourcesAI companies accused of destroying rare books through destructive scanning
Booksellers report a surge in massive, unusual orders for books suspected to be part of ‘destructive scanning’ by AI companies to rapidly digitize content for training models.
-
[TECHNOLOGY] 3 sourcesAI labs shred millions of books to train models
AI labs, including Anthropic’s “Project Panama,” are buying, scanning and shredding millions of pre‑2022 books for training data, prompting legal rulings on fair use and concerns from Australian booksellers.
-
[TECHNOLOGY] 8 sourcesElon Musk Orders SpaceXAI to Scan Rare Books Without Destroying Them
Elon Musk told SpaceXAI to use non‑destructive scanners for rare books, after reports that Anthropic ripped spines and destroyed originals while training its Claude AI model.
-
[TECHNOLOGY] 27 sourcesAnthropic and Other AI Firms Scan and Destroy Millions of Old Books
AI firms are bulk‑buying pre‑2022 printed books, using destructive scanning that cuts spines and destroys originals. Anthropic’s “Project Panama” targeted up to two million books; a U.S. judge ruled such scans
Sources
919thebend.ca · adevarul.ro · agorarn.com.br · americanbazaaronline.com · apjii.or.id · bitcoinethereumnews.com · botosaninews.ro · bpmoney.com.br · businessmag.al · christianitytoday.com · crypto-times.jp · dailynews.co.tz · dailytimes.com.pk · decrypt.co · digitalphablet.com · digitaltrends.com · english.alsiasi.com · finance.technews.tw · freerepublic.com · fultonsports.com · gazetezebra.com.tr · gizmodo.com.au · gizmodo.jp · gomag.com · hotnews.ro · huffingtonpost.it · ibtimes.co.uk · ihedn.fr · it-boltwise.de · ithome.com · karar.com · karlismyunkle.com · larissapress.gr · letemsvetemapplem.eu · me.mashable.com · merca2.es · new.americanprophet.org · news20.ro · newswithviews.com · nuevarevista.net · openculture.com · pedestrian.tv · perlentaucher.de · physicalers.com · rayhaber.com · realtid.se · russh.com · searchenginejournal.com
This summary has been updated 11 times: see revision history