What Is Destructive A.I. Scanning, And Why Has It Destroyed Millions of Books?

The investigations began from some curious independent media writers and reporters who noticed a number of tech development and other firms were buying up HUGE quantities of books.  Book curators and sellers also noticed a large increase in orders from these companies.

  Amazon Has Been Practicing Destructive Scanning for at Least Several Years

One of the ways tech companies 'train' their AI systems is to feed them data. The more accurate information, the 'smarter' the programs become.  This involves troves of information from books, among other sources.

Get our free mobile app

Companies target works published before 2022, because they're less likely to contain any 'tainted' AI content.  404 Media began to expose a lot of this when they started to investigate some large book orders from Amazon, beginning with a company that placed the order for 1,000 books through Biblio.

  Biblio deals online with rare and used books

Curious to see where the books were headed, and working with the book vendor, 404 slipped an Apple Air Tag tracker into one of the books ,and they wound up at an Amazon Las Vegas warehouse where workers were ripping off the spines, scanning the pages, and then destroying the books.  Inc.com online also said Amazon was not the only one:

"Amazon is not alone. Court records revealed that Anthropic bought millions of print books, cut off their bindings, scanned them, and then destroyed the originals while building a digital research library."

Anthropic had been sued over the practice. They were accused of copyright infringement, but a Judge ruled their practice of creating a digital copy of each book they destroy is legal--as long as multiple copies were not made.

  Why Are the Spines Cut Off and Books Destroyed?

It's a matter of cost and time.  It takes far too long to turn a page and scan it, as opposed to ripping off the spine, feeding the pages into a scanner, and then into the AI system.

The bottom line is, a company buys the books, and what it does with them is their choice, and the practice--although seemingly barbaric--is apparently legal. But it calls into question how many possibly rare and even priceless books have slipped through and been destroyed forever.

It's estimated the number of books destroyed for the purpose of training AI is well into the tens of millions.