Anthropic Settlement List Includes 21 Korean Books

As of October 17, 2026, the Literature Translation Institute of Korea says it found 21 books by Korean authors on the list of titles covered by U.S. AI company Anthropic's copyright settlement. English editions by Han Kang, Hwang Sok-yong and others are included, and the institute has told the authors what happened and how to seek compensation.

21 booksKorean-authored titles on the settlement list

About 500,000 worksWorks the court recognized as infringed

About $1.5 billion (roughly 2 trillion won)Total settlement amount

About $3,000Compensation per work

Key points

The Literature Translation Institute of Korea found 21 Korean-authored titles among the books eligible for compensation on the settlement website for the Anthropic lawsuit.
The list includes works by Han Kang, Hwang Sok-yong, Kim Young-ha, Shin Kyung-sook, Kim Ae-ran, Kim Soom and Hwang Sun-mi.
A U.S. court recognized infringement of about 500,000 works, and the case ended with a settlement in July.
Korean authors can receive roughly $3,000 per work if they take part in the formal claims process.

While going through the list of compensation-eligible books posted on the copyright settlement website for the Anthropic lawsuit, the Literature Translation Institute of Korea found 21 works by Korean authors. The institute has since notified the authors and explained the damage and their options for relief. Authors on the list include Han Kang, Hwang Sok-yong, Kim Young-ha, Shin Kyung-sook, Kim Ae-ran, Kim Soom and Hwang Sun-mi. The titles are English translations of novels such as "Please Look After Mom," "Your Republic Is Calling You" and "The Hen Who Dreamed She Could Fly."

Anthropic faced a class action in the U.S. from authors and publishers over allegations that it illegally downloaded copyrighted works while training its chatbot, Claude. The court recognized infringement of roughly 500,000 works, and the case ended in July when Anthropic agreed to pay a settlement. The total is about $1.5 billion, or around 2 trillion won. That works out to roughly $3,000 per work.

According to some reports, the U.S. court found that using books to train AI is itself fair use under copyright law. But it did not extend that protection to the pirated copies Anthropic downloaded from piracy sites and kept to build its training library. The litigation also showed that Anthropic downloaded large volumes of e-book files from pirate sites.

The institute says it will work with copyright organizations to help authors get fair compensation, and will support legal action if needed. It is reportedly preparing detailed measures to help with claims. To actually get paid, authors must take part in the formal claims process themselves.

In Korea right nowKorea's publishing industry is taking a harder line on AI training and copyright after the finding. On October 16, the Korean Publishers Council issued an appeal to the government, the National Assembly and AI companies. It criticized a reality in which the people who make books can't even find out which AI used their work, or how. The group demanded data disclosure, prior permission from rights holders and fair compensation. Reports say 163 publishers and organizations and 333 authors, translators and publishers signed on. Controversy grew after reports that Woongjin Booxen, a Korean book distributor, signed a supply deal with Anthropic in 2024 and delivered books the following year, and that an overseas used-book buyer pitched large bulk orders to major bookstores.

FAQ

Why are Korean authors' books on a U.S. settlement list?

According to the institute, Korean literary works that were translated, including into English, and sold abroad were among those Anthropic illegally downloaded. Once those books were listed as eligible for compensation, the harm to Korean authors was confirmed.

How much can Korean authors receive?

Compensation is reported at about $3,000 per work. Actually receiving it depends on the author taking part in the formal claims process.

Did the court rule that training AI on books is illegal?

Not according to reports. The court treated the training use as fair use and faulted only the downloading and storing of pirated copies.