Anthropic destroyed millions of print books to build its AI models

On Monday, court documents revealed that AI company Anthropic spent millions of dollars physically scanning print books to build Claude, an AI assistant similar to ChatGPT. In the process, the company cut millions of print books from their bindings, scanned them into digital files, and threw away the originals solely for the purpose of training AI—details buried in a copyright ruling on fair use whose broader fair use implications we reported yesterday.

The 32-page legal decision tells the story of how, in February 2024, the company hired Tom Turvey, the former head of partnerships for the Google Books book-scanning project, and tasked him with obtaining "all the books in the world." The strategic hire appears to have been designed to replicate Google's legally successful book digitization approach—the same scanning operation that survived copyright challenges and established key fair use precedents.

While destructive scanning is a common practice among some book digitizing operations, Anthropic's approach was somewhat unusual due to its documented massive scale. By contrast, the Google Books project largely used a patented non-destructive camera process to scan millions of books borrowed from libraries and later returned. For Anthropic, the faster speed and lower cost of the destructive process appears to have trumped any need for preserving the physical books themselves, hinting at the need for a cheap and easy solution in a highly competitive industry.

Ultimately, Judge William Alsup ruled that this destructive scanning operation qualified as fair use—but only because Anthropic had legally purchased the books first, destroyed each print copy after scanning, and kept the digital files internally rather than distributing them. The judge compared the process to "conserv[ing] space" through format conversion and found it transformative. Had Anthropic stuck to this approach from the beginning, it might have achieved the first legally sanctioned case of AI fair use. Instead, the company's earlier piracy undermined its position.

But if you're not intimately familiar with the AI industry and copyright, you might wonder: Why would a company spend millions of dollars on books to destroy them? Behind these odd legal maneuvers lies a more fundamental driver: the AI industry's insatiable hunger for high-quality text.

The race for high-quality training data

To understand why Anthropic would want to scan millions of books, it's important to know that AI researchers build large language models (LLMs) like those that power ChatGPT and Claude by feeding billions of words into a neural network. During training, the AI system processes the text repeatedly, building statistical relationships between words and concepts in the process.

The quality of training data fed into the neural network directly impacts the resulting AI model's capabilities. Models trained on well-edited books and articles tend to produce more coherent, accurate responses than those trained on lower-quality text like random YouTube comments.

Publishers legally control content that AI companies desperately want, but AI companies don't always want to negotiate a license. The first-sale doctrine offered a workaround: Once you buy a physical book, you can do what you want with that copy—including destroy it. That meant buying physical books offered a legal workaround.

And yet buying things is expensive, even if it is legal. So like many AI companies before it, Anthropic initially chose the quick and easy path. In the quest for high-quality training data, the court filing states, Anthropic first chose to amass digitized versions of pirated books to avoid what CEO Dario Amodei called "legal/practice/business slog"—the complex licensing negotiations with publishers. But by 2024, Anthropic had become "not so gung ho about" using pirated ebooks "for legal reasons" and needed a safer source.