Anthropic Pays $1.5B in Largest AI Book Copyright Settlement

Anthropic agreed in September to pay $1.5 billion to settle claims that it trained its Claude chatbot on pirated books, the largest copyright settlement in US history. The deal covers more than 482,000 titles and sets a payout of $3,000 per illegally downloaded book. A federal court had already ruled that Anthropic's use of those pirated copies was unlawful, though training on legally purchased books still counts as fair use.

Now that checks are supposed to go out, the settlement has turned into a scramble over who actually owns the money. The New York Times reports that authors and publishers are filing competing claims with the settlement administrator, each side arguing it holds the rights to specific titles. Mary Rasenberger of the Authors Guild says the mess stems partly from publishers who never bothered to track which rights had reverted back to authors after contracts lapsed.

Textbook writers appear to be losing the most ground, with some contracts leaving them a fraction of their share. Literary agencies have jumped in too, staking claims to money they have no legal right to collect, according to Writer Beware. Where publishers and authors can't settle it themselves, a court-appointed arbitrator will make the call.