
What happened
Nippan Group Holdings told the Japan Book Publishers Association on the 5th that its subsidiary sold Japanese books, including titles from member publishers, to Anthropic between December 2024 and December 2025.
Why it matters
The admission confirms that Japanese publishers' titles reached U.S. AI training pipelines through a major distributor, potentially exposing them to copyright disputes they did not authorize.
What to watch
Nippan says it did not recognize the sales were meant for AI training, so the test is whether the association accepts that defense or presses for damages and tighter sales controls.
WHO IT HITSJapanese book publishers and rights holders whose titles moved through Nippan's distribution network now face questions about whether their works were used for AI training without their consent or knowledge.
Summaries like this, in your inbox every morning.
The disclosure places Nippan, one of Japan's largest book distributors, in the middle of a widening global conflict over whether copyrighted works can be used to train generative AI. Nippan has acknowledged the sales occurred over roughly a year and has said it did not recognize that AI training was the purpose. That distinction is now central, because the Japan Book Publishers Association asked about the sales and received a formal answer dated the 5th.
The case also follows evidence that Anthropic has faced copyright litigation in the United States, including a settlement approved by a U.S. court and a lawsuit from Sony Group's music publishing arm over unauthorized training. Those disputes show that rights holders are increasingly willing to challenge AI companies and their suppliers in court, not just in public statements.
For publishers, the immediate question is whether Nippan's lack of awareness will be treated as a sufficient defense or as evidence that distribution channels need stronger safeguards. For distributors and AI developers, the outcome could hinge on whether bulk sales of copyrighted material require explicit disclosure of AI training as the end use.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
NetApp's Jen Prenner said legacy data can be made AI-usable without re-architecting it, and NetApp announced p…
Taiwan's National Science and Technology Council (NSTC) launched two research programs targeting silicon photo…

SpaceX launched a prototype satellite for Alphabet's Project Suncatcher carrying four Tensor Processing Units…

Mistral AI CEO Arthur Mensch unveiled Mistral Large 4 (ML4) at AI Everything Abu Dhabi, a 1.05 trillion-parame…

Vercel COO Jeanne DeWitt Grosser said the company's inbound sales agent evolved from a 1,000-line prompt to 14…

Mistral released a public preview of Mistral Large 4, a trillion-parameter model with 49 billion active parame…
