Artificial intelligence firms are purchasing millions of second-hand books from international sellers to fuel their training datasets. These physical copies are being scanned and subsequently destroyed, sparking a fierce debate over copyright theft and the erasure of cultural history.
Anthropic's Project Panama and the Millions of Destroyed Pages
Internal documents have revealed a clandestine operation known as Project Panama, conducted by the AI firm Anthropic. According to the report, Anthropic spent millions of dollars since early 2024 to acquire millions of print books to expand its training databases beyond the limits of the searchable internet. This initiative was designed with a strict mandate for secrecy, as internal memos indicate that employees were explicitly told to keep the project hidden from the public.
The process employed by Anthropic and other tech giants is described as "destructive scanning." In this workflow, books are transported to warehouses in the United States where their pages are sliced away from the spines. These pages are then fed into high-speed industrial scanners to harvest text for models like Claude and ChatGPT before the original paper copies are discarded. This method allows AI companies to digitize vast amounts of niche information quickly and at a low cost.
From Compass Books in Devon to Heathrow Shipping Hubs
The phhysical sourcing of this data has hit independent bookstores in Britain, such as Compass Books in Dartmouth, Devon. The owner, Lucy Hingston, noticed an unusual surge in bulk orders for obscure titles, ranging from the geology of western Peru and the history of Cornish pioneers in South Africa to the specifics of jewellery in Hong Kong and global cheese varieties. As the report says, Hingston became suspicious when she discovered the buyer was actually a shipping company based at Heathrow airport rather than a traditional collector or researcher.
This pattern sugests a coordinated effort to pillage British bookshops for materials that have never been uploaded online. By targeting pre-internet era materials and niche subjects, Silicon Valley firms aim to give their AI responses more depth and nuance, making the output feel more human. However, for sellers like Lucy Hingston, this represents a moral tragedy where curated knowledge is treated as raw material for an "AI monster."
Dan Conway and the 'Brazen Theft' of Creative Labor
The legal ramifications of these practices are severe, with Dan Conway, the CEO of the Publishers Association, condemning the actions as "brazen theft." Conway argues that using content from books and journals without payment or permission is a direct attack on the livelihoods of journalists, illustrators, and authors. This shift from digital scraping to physical destruction marks a more aggressive phase of the AI gold rush, where intellectual property laws are bypassed to secure a competitive advantage.
In response to this systemic exploitation, the Make It Fair campaign has emerged to protect creative works from being absorbed by big tech. This movement has gained high-profile support from figures such as Simon Cowell and Sir Elton John . The campaign highlights a growing tension between the tech industry's need for incomprehensible amounts of text to predict the next token in a sequence and the rights of the humans who created that text.
Which Rare Editions Have Already Vanished?
While the scale of Project Panama is documented in court records, several critical questions remain. The source mentions that the "last remaining editions of historical tomes" are being permanently erased, but it does not specify which titles or rare collections have already been destroyed. Furthermore, while Anthropic is named specifically, the report refers to "tech giants" in the plural, leaving it unclear which other AI firms are utilizing Heathrow-based shipping hubs for destructive scanning.
There is also a lack of clarity regarding the legal jurisdiction of these acts; while the books are sourced in Britain, the destruction occurs in the United States. It remains to be seen whether the Publishers Association or the Make It Fair campaign can successfully litigate against the physical destruction of copyrighted materials once they have left UK soil.
Comments 0