Authors contest third-party claims on Anthropic's $1.5 billion payout

Authors expecting payouts from Anthropic's $1.5 billion copyright settlement received emails this week saying publishers and literary agents had filed competing claims on their shares, forcing a federal court to referee who gets paid from the AI-training fund.

The accord, reached in September 2025, ended a class action over pirated training data after a judge found that training AI on copyrighted books qualified as fair use but that obtaining the works through piracy did not. Anthropic has said it disagrees with the publishers' claims and intends to defend itself in court, language it used last week in response to a separate music-publisher lawsuit.

The distribution fight comes as the copyright battle widens. Sony Music Publishing and Warner Chappell Music sued Anthropic in California federal court last Friday, alleging the company trained its Claude models on tens of thousands of copyrighted songs obtained through torrenting and scraping, and seeking statutory damages of up to $150,000 per infringed composition. The complaint describes the conduct as "one of the largest and most blatant ongoing thefts of intellectual property in history."

How the court divides the $1.5 billion will set a template for future AI training-data settlements, a question that carries weight as Anthropic weighs an initial public offering. The company disclosed preliminary second-quarter revenue above $11.5 billion, more than 14 times the year-earlier level, and its first positive adjusted operating income.

The emails to authors mark the first visible friction in the payout phase of a settlement that took shape over more than a year. In June 2025 a judge determined Anthropic had acquired more than 7 million unlicensed books from the digital repositories Library Genesis and Pirate Library Mirror, a finding that underpinned the authors' class action even after the same court concluded that using the copyrighted literature to train models was fair use.

The competing claims pit individual authors against the publishers and agents who hold or administer rights to the same works, a structural tension that has defined AI content disputes since the first wave of litigation. Universal Music Group, Concord Music Group and ABKCO won court-sanctioned restrictions in 2023 requiring Anthropic to stop Claude from reproducing copyrighted lyrics, an early precedent that music publishers now cite as they press for damages.

The stakes extend beyond the authors in this case. A ruling that publishers and agents can carve out shares of individual authors' settlement payments would reshape how future AI training-data deals are structured, steering compensation toward rights holders rather than creators. That allocation matters to the broader technology sector, where Anthropic's revenue run-rate and its path toward a listing make the final cost of its copyright exposure a material line item for investors.

Anthropic co-founders Dario Amodei and Benjamin Mann were named alongside the company as defendants in the music-publisher suit, which cites songs including Mariah Carey's "All I Want for Christmas is You," Taylor Swift's "Paper Rings" and Survivor's "Eye of the Tiger." The publishers contend Claude can reproduce near-identical copies of compositions and generate lyrics that compete with human-made songs, threatening the licensed-lyrics market.

The court has not yet set a schedule for resolving the competing claims on the settlement fund. Authors who received the emails this week have begun pushing back against the third-party filings, setting up a procedural fight that could delay payouts while the judge weighs the validity of each claim. If the court sides with the publishers and agents, future settlements may route compensation through rights holders; if it backs the authors, creators keep a larger share of any AI-training payout.

This article is for informational purposes only and does not constitute investment advice.