Huaxia Publishing House Labels Books to Prohibit AI Training
Huaxia Publishing House has begun explicitly labeling its books to prohibit their use in AI training. While acknowledging the difficulty of legal enforcement, the publisher aims to assert copyright awareness in response to AI companies scraping digitized books. This move highlights the growing conflict between traditional publishers and AI developers over copyright and data scraping. It represents a symbolic step for content creators trying to protect their intellectual property as AI companies increasingly turn to digitized books for training data. The publisher decided to add this warning to many of its non-imported books after internal discussions about AI and large language models. This action follows public outcry over AI companies purchasing physical books and using destructive scanning methods—cutting off book spines for high-speed scanning—to quickly digitize training materials.
## BACKGROUND
Destructive book scanning involves cutting the spine off a book to feed individual pages into high-speed scanners, which is faster and cheaper than non-destructive methods but destroys the physical book. AI companies have reportedly used this method to digitize massive volumes of text to train large language models, raising significant copyright and ethical concerns among authors and publishers.