Korean Authors Receive Anthropic Settlement Payouts
Twenty-one book titles qualify for compensation following a $1.5 billion legal settlement regarding AI training data.
Updated on Oct. 21, 2026 in Artificial Intelligence
Live Poll
Should authors receive mandatory compensation when their books are used to train artificial intelligence?
Following a $1.5 billion settlement approved in July 2026, 21 books by Korean authors have been deemed eligible for $3,000 in compensation each. The legal resolution follows class-action litigation in the United States regarding the unauthorized use of copyrighted works in AI model training.
Why it matters
The case highlights the growing friction between AI companies and the publishing industry over data scraping, as authors and publishers now seek to track bulk book purchases potentially intended for AI dataset ingestion. This shift forces booksellers to scrutinize large-scale orders to prevent intellectual property exploitation.
The settlement covers 21 confirmed book titles from Korean authors, each qualifying for a $3,000 payout. These figures were established following a federal court ruling in July 2026 that addressed the training of generative models on protected literary data.
The players
Anthropic PBC
An AI research and safety company that builds large language models and currently faces legal challenges over its training data sourcing.
Literature Translation Institute of Korea
A government-affiliated organization that supports the global dissemination of Korean literature and is currently assisting authors with copyright claims.
Kyobo Book Centre
A major South Korean bookstore chain that has reported receiving bulk purchase requests potentially linked to AI training efforts.
Korean Publishers Association
An industry group representing publishing interests that is currently surveying the market to identify unauthorized AI training data harvesting.
The details
The dispute centers on how Anthropic utilized large-scale datasets, which plaintiffs alleged included pirated copyrighted materials. In response, institutions like the Literature Translation Institute of Korea cross-referenced settlement lists with local databases to identify impacted authors. Concurrently, bookstores including Kyobo Book Centre and Yes24 have faced suspicious requests from overseas distributors for bulk orders of 100,000 titles, leading to order denials due to concerns that these books were intended to fuel future AI training cycles.
Timeline
2024: Anthropic was sued in a class-action lawsuit.
Early 2026: Kyobo Book Centre began receiving bulk purchase requests from ten overseas companies.
July 2026: A federal judge granted final approval of the $1.5 billion settlement.
Late August 2026: A US distributor contacted Yes24 for a bulk purchase of used books.
The Tech Race
This development serves as a downstream effect of the legal precedent set by Bartz, et al. v. Anthropic PBC. It marks a shift from broad class-action litigation toward the targeted identification and compensation of individual international rights holders impacted by AI training.
Authors and publishers are now tasked with cross-referencing their catalogs against settlement databases to verify if their works are eligible for payouts. Meanwhile, booksellers are implementing stricter screening processes for bulk orders to mitigate the risk of their inventory being repurposed for AI model development.
The takeaway
The Korean Publishers Association is forming a new coalition to monitor and challenge the mass acquisition of published works for AI model training. Industry observers should watch for how these regional organizations coordinate future legal strategies to ensure compensation for creators.
Further reading
For broader trends in model training and legal standards, see the Artificial Intelligence section.
Live Poll
Should authors receive mandatory compensation when their books are used to train artificial intelligence?
