Z-Library’s Pirated Books Fuel AI Training Race

Shadow libraries such as Z-Library and Library Genesis, or LibGen, spent years assembling unauthorized copies of books and academic papers, widening access while drawing sustained legal attacks from publishers and governments. Their collections also became unusually valuable to artificial-intelligence developers: books offer coherent, edited, long-form language that can outperform noisy web text as training material for large language models. Because the files circulate through torrents and mirrors, taking down a website does not erase the underlying corpus.

The U.S. Justice Department seized Z-Library domains and charged two alleged Russian operators in November 2022. Days later, Anna’s Archive launched to index copies from Z-Library, LibGen and other repositories. A May 5, 2026, publisher complaint said Meta obtained more than 81 terabytes through Anna’s Archive; separate court findings showed Anthropic downloaded more than 7 million pirated books. Anthropic agreed in September 2025 to a $1.5 billion settlement covering roughly 500,000 works, underscoring how shadow-library files survived enforcement and became strategic AI inputs.