In stealing our voice, Meta has stolen part of our essence

Over the last few days, you may have seen images posted by your favourite authors of their books listed on the Library Genesis LibGen) shadow library”. While libraries in general are excellent, LibGen is no such thing. It is a pirate site that has taken our books and put them online for people to read for free.

- ABIE LONGSTAFF

At the Society of Authors, we have long suspected that these pirate sites are being used to train AI machines. The machines, called large language models or LLMs, are fed millions and millions of texts, which are then used to create content on AI sites like ChatGPT, Claude and others.

It has recently emerged that Meta, the social media giant behind Facebook, WhatsApp and Instagram, scraped our books from LibGen, where they were already illegally hosted. Millions of books have been taken. Your favourite authors are very likely represented there. Meta argued that it made “fair use” of the books in developing its LLM, LLama, arguing that a class action suit brought by authors including Ta-Nehisi Coates and Sarah Silverman should be thrown out. Court documents show that Meta considered signing licensing agreements with authors and publishers but decided the process would be “incredibly slow” and “unreasonably expensive”, so instead, it just stole our work.