Rare Books Dismantled and Destroyed to Train Artificial Intelligence Models, Investigation Reveals

In what is being described as an alarming intersection of technological expansion and cultural erosion, thousands and potentially millions of books are being systematically acquired, physically dismantled, and scanned to fuel the development of artificial intelligence systems. This opaque operation, which treats physical texts as mere raw material for computational models, was detailed in an investigation led by Emanuel Maiberg, an investigative journalist and co-founder of the independent media outlet 404 Media.

Speaking in an interview on France 24’s program "Spotlight," hosted by François Picard, Maiberg outlined the mechanics of an unexpected and deeply troubling pipeline. As the artificial intelligence sector pushes into uncharted territory, its voracious appetite for high-quality textual training data has driven a global procurement effort targeting rare and physical volumes, effectively turning cultural artifacts into digitized parameters.

According to Maiberg's reporting, this acquisition pipeline operates through a deliberately secretive and international supply chain. Intermediaries and buyers utilize anonymous online marketplaces to amass massive quantities of print material without disclosing the ultimate destination or purpose of their purchases. This built-in opacity allows bulk buyers to gather vast inventories of physical literature quietly, away from the scrutiny of the public and traditional cultural institutions.

Once procured, these volumes are transported to large-scale industrial scanning facilities where the process takes a starkly destructive turn. Rather than employing non-invasive archival scanning methods, workers at these facilities physically slice away the spines and bindings of the books. Stripping the bindings allows the dismembered, individual pages to be fed at rapid speeds through high-capacity scanners, maximizing data intake while destroying the physical book itself.

The scale of this operation is extensive, spanning thousands if not millions of individual works. The sheer speed at which artificial intelligence models require fresh, sophisticated training data has transformed what was once a careful, scholarly digitization process into an industrial assembly line, colloquially likened to an AI woodchipper that permanently consumes the printed word.

Maiberg emphasized that the implications of this practice extend far beyond the striking and dystopian visual of books being torn apart. At its core, the mass destruction represents a fundamental shift in how human knowledge is stored, owned, and accessed. For centuries, a physical book has functioned as an open-ended, enduring vessel of thought, capable of circulating freely among private readers, public libraries, academic institutions, and future generations.

By contrast, once a physical text is dismantled, scanned, and absorbed into an artificial intelligence architecture, its fundamental nature is altered. The text survives only virtually, broken down and encoded within a proprietary neural network. In this new format, the knowledge is no longer governed by public lending rights or shared cultural commons; instead, its outputs and rules of access are strictly dictated by the private corporation that owns the model.

This dynamic exposes a profound and unsettling paradox at the heart of the current artificial intelligence boom. The very industry that claims to be cataloging and synthesizing the vast breadth of human knowledge is simultaneously responsible for destroying the tangible, independent sources from which that knowledge originates, leaving society with derivative digital models rather than authentic physical records.

Crucially, Maiberg noted that this systematic destruction is not animated by a deliberate or malicious campaign to rid society of printed literature. Rather, the driving force behind the practice is something arguably more disturbing: absolute institutional indifference. In the frantic commercial race to construct increasingly capable AI models, physical books are viewed merely as algorithmic inputs waiting to be optimized.

Within this efficiency-driven calculus, the physical destruction of rare and historical books is not seen as a cultural tragedy, but simply as an acceptable and negligible operational cost. The technical speed gained by severing a book's binding outweighs any value placed on the long-term preservation of the physical artifact, highlighting a mindset where tangible heritage is subordinated to digital processing speed.

Maiberg argued that this operational logic is part of a broader historical trajectory concerning the concentration of informational power. Over the past several decades, the rise of dominant search engines and centralized social media platforms demonstrated how private technology conglomerates could monopolize the distribution of information, gradually positioning themselves as indispensable gatekeepers of public discourse.

However, the rapid advancement of artificial intelligence is now taking that concentration of power to an unprecedented level. Rather than merely indexing web pages or hosting digital conversations, AI corporations are directly absorbing human cultural expression and synthesizing it into closed systems, fundamentally changing the relationship between creators, readers, and information.

The findings uncovered in the investigation pose urgent questions for the worlds of culture, literature, and academia. As increasing amounts of humanity’s accumulated knowledge become mediated almost exclusively through artificial intelligence platforms, serious concerns are emerging regarding who will own and operate the foundational infrastructure of human thought.

Scholars and cultural observers are left to confront a reality where access to historical knowledge may increasingly require interacting with proprietary AI platforms that dictate terms of use, pricing, and content delivery. The transition from public, shelf-stable books to corporate-controlled digital models threatens to redefine how future generations interact with history and literature.

Ultimately, Maiberg’s revelations on "Spotlight" illustrate the hidden costs beneath the rapid expansion of artificial intelligence. As the tech sector pushes forward, the mass destruction of physical books serves as a stark reminder of how public knowledge is quietly being transformed into private intellectual property under the banner of progress.