Anthropic’s Project Panama Raises Questions Over AI Training and Book Destruction

Anthropic’s Project Panama has sparked debate after reports revealed millions of printed books were dismantled and scanned to train Claude AI, raising concerns over copyright, ethics, and book preservation.
on Jul 30, 2026

Anthropic’s “Project Panama” Fuels Controversy Surrounding AI Training Data and Books Preservations

The recent controversies surrounding Anthropic were sparked by the release of the information regarding their Project Panama, which revealed that Anthropic had physically broken down millions of printed books and used them to train the Claude AI models created by the company. This information was released by court documents in a copyright infringement lawsuit against Anthropic.

According to sources, Anthropic purchased many used books, stripped off their covers, scanned each page of the books using industrial scanners, and then destroyed the paper copy of those books. This scanning process called “destructive scanning” was chosen by Anthropic, since it proved to be much faster and cheaper than the alternative process of scanning.

Tom Turvey, who used to work on Google Books, allegedly headed this project. He wanted to create a huge database of high-quality texts written by humans and not created by AI in order to train their language models. With increasing use of AI-written content on the Internet, companies seek good sources of pre-AI literature.

Although the process may be deemed acceptable legally because of fair use laws in some cases, it has faced condemnation by writers, publishers, and organizations concerned with preservation. The concern is that the destruction of physical books, especially rare or out-of-print versions, would undermine cultural heritage and allow valuable literature to be stored privately by AI technology that cannot be accessed publicly.

There are also concerns raised in relation to copyright laws, compensation of artists and moral obligations of AI companies. Although the courts have separated legally purchased physical copies of books from the illegal digital collection that was being disputed, the issues raised by the exposure of Project Panama have brought up the need for regulations around AI training data.