In industrial warehouses, special machines would cut off the spines of books, and then their content was digitized en masse. Within milliseconds, the text would enter the digital space and be used to train artificial intelligence models.
Read more Prank on Russian officials: Ukrainian intruded into a secret conversation and caused panic
Later, the paper would become waste, from which toilet paper and cardboard were made. Internal documents of the “Panama” project reveal that Anthropic’s plan was to “destructively digitize all the world’s books” and to do all this secretly so that no one would find out.
What is the “Panama” project?
The “Panama” project was started in early 2024 when Anthropic’s management was looking for books that could teach their AI assistant Claude to “write well,” rather than copy “low-quality internet language.”
The company was buying tens of thousands of books at a time from several suppliers, including Better World Books and the UK-based World of Books.
According to the contractor who carried out the book digitization and destruction work, Anthropic was looking for an “experienced document scanning service provider who could digitize between 500,000 and two million books within six months.”
Anthropic’s management was convinced that by training Claude from books, rather than internet texts already contaminated with AI-generated content, this AI program could be taught to write much more smoothly.
According to one document, one of the company’s founders, Ben Mann, received a huge collection of pirated books from the “shadow library” LibGen in June 2021. The following year, he shared a link to the illegally obtained book catalog “Pirate Library Mirror” with his colleagues.
Read more Do you think that 10,000 steps a day is a necessity? Revealed how many you really need to walk
What is the legal resolution?
The existence of the Panama project became known only from thousands of pages of court documents that were made public during a copyright infringement lawsuit filed by a group of writers claiming that Anthropic illegally used their works to train Claude.
In 2024, a group of authors led by writer Andrea Bartz, journalists, and book authors Charles Graeber and Kirk W. Johnson filed a class-action lawsuit against Anthropic, claiming the company violated copyright laws by using pirated books to train its AI model.
In July of this year, Anthropic, valued at $965 billion, agreed to pay $1.5 billion to settle the case involving about 500,000 works. “This is the largest known compensation in copyright cases in history,” said lead plaintiffs’ attorney Justin Nelson.
This case became the first of many lawsuits against AI companies to be resolved peacefully.
Meta, OpenAI, Google, and Microsoft are also facing copyright infringement lawsuits filed by authors making similar accusations of piracy of their works.
Read more Pope Leo XIV will meet with victims of clergy abuse in France
