Japan Publishers Challenged AI Book Sales
The Japan Book Publishers Association has formally demanded answers from distributor Nippan Group Holdings regarding AI training.
Updated on Sept. 21, 2026 in Artificial Intelligence
Live Poll
Should AI companies be required to obtain permission before using copyrighted books for training?
The Japan Book Publishers Association, representing 380 member companies, has requested an explanation from Nippan Group Holdings over suspected book sales to the AI firm Anthropic. This inquiry follows internal Anthropic documents that suggest a sales contract was established in November 2024.
Why it matters
Publishers are deeply concerned that unauthorized book usage for AI learning purposes constitutes significant copyright infringement. This action highlights the growing friction between content creators and technology firms over data sourcing for model training.
Anthropic allegedly utilized a website for pirated copies to facilitate its AI learning processes. Internal files cite a potential sales agreement between the AI developer and Nippan signed in November 2024.
The players
Japan Book Publishers Association
This trade organization represents 380 member companies and serves as a representative body for the Japanese publishing industry.
Nippan Group Holdings
This company operates as a major book distributor and has denied claims that it sold publications to artificial intelligence firms.
Anthropic
This artificial intelligence company is the subject of claims regarding the acquisition of copyrighted works for model training.
The details
The association has requested that Nippan Group Holdings provide a detailed list of all books sold and clarify the legal basis for these transactions. While the association continues to oppose the use of books for AI model training, Nippan has explicitly denied selling any works to AI companies for such purposes.
Timeline
November 2024: Anthropic allegedly signed a sales contract with Nippan.
Late July 2026: Anthropic settled a lawsuit with American writers.
September 21, 2026: The inquiry regarding the alleged sales was publicized.
The Tech Race
The current inquiry into Japanese distribution practices follows the pattern set by the July 2026 settlement between Anthropic and American writers regarding copyright usage. This signifies an escalation in the ongoing global effort by rights holders to police the data sets powering generative AI.
This development could eventually alter how AI models are trained, potentially resulting in different capabilities or access to information in consumer-facing tools. Users may experience changes in the availability of specific AI services as companies navigate these shifting copyright requirements.
The takeaway
The tension between publishers and AI companies suggests that copyright enforcement will continue to shape the development of new technologies. Stakeholders should monitor future disclosure requirements as they may set new standards for transparency in data sourcing.
Further reading
For more on the industry's response to these technologies, visit Artificial Intelligence.
Source note: This article includes information reported by 朝日新聞デジタル.
Live Poll
Should AI companies be required to obtain permission before using copyrighted books for training?
