AI training and copyright challenges impact global publishing

The publishing industry is facing significant challenges regarding the use of artificial intelligence in content creation and data training. In Japan, court documents have revealed that the major book distributor Nippan entered into a contract with the US-based AI startup Anthropic to sell large quantities of books. These books are reportedly being shredded and scanned to create electronic datasets for AI training purposes.

Simultaneously, the industry is grappling with the difficulty of verifying human authorship. Publishers and literary agents are increasingly using AI detection tools like Pangram to vet manuscripts, leading to a ‘crisis of subjectivity’ where authors may feel pressured to provide extensive proof of their creative process to avoid being flagged. The uncertainty surrounding whether AI-generated content can be protected by copyright remains a major commercial risk for publishers and creators.