Games Workshop is famously anti-AI, but I was still caught by surprise when the narrator of my new favorite Warhammer 40,000 audiobook read out a warning about training AI systems on the novel itself.
AI models training on books without permission has become one of the big topics in the tech space. Last week, a report by
404 Media
alleged that Amazon was buying “massive quantities of books, scanning them for AI training data, and destroying them in the process.”
Last summer, it emerged that Anthropic had shredded millions of physical books to train its Claude AI model. The secret initiative, called Project Panama, was unearthed in a lawsuit brought by a group of authors against Anthropic, which eventually agreed to settle for $1.5 billion despite the ruling largely siding with the AI company on fair use.
And in October last year, a federal court advanced two new theories of infringement against ChatGPT-owner OpenAI in a lawsuit brought by authors including Game of Thrones’ George R.R. Martin. No final summary judgment ruling has been issued yet.
And there is evidence to suggest AI models are trained on Warhammer books. In March 2025, an investigation by The Atlantic found that Meta allegedly pirated millions of books to train its AI. Black Library authors were among those subsequently horrified to discover that their novels were in LibGen
, the pirated-books database that Meta used to train AI. A quick search shows pretty much all of Robert Rath's Warhammer novels are in there.
Meanwhile, fake AI-generated Warhammer books have become
a real problem
, with unlicensed Warhammer 40,000 books complete with clearly AI-generated covers and apparent AI-generated text available on various retailers.