Authors Push Back as Publishers Claim Share of Anthropic's $1.5 Billion Settlement

A dispute has emerged between authors and publishers over the distribution of Anthropic’s $1.5 billion copyright settlement. According to reporting by Anthony Ha of TechCrunch AI, authors who were supposed to receive payments have been notified that third parties have filed claims. The large-scale settlement covers about 500,000 books, with $3,000 to be paid per work. Final approval was granted in July, and payment procedures are now underway.

The settlement stems from a class-action lawsuit over the training of AI models. The court ruled that the act of using copyrighted works for training itself was lawful under the fair use doctrine. On the other hand, it found that acquiring training data via pirated copies was unlawful. Following that ruling, Anthropic agreed last year to settle for $1.5 billion. The case is notable for distinguishing between the legality of AI training and the illegality of data acquisition. A similar pattern is spreading to lawsuits against other AI developers. In Round Hill’s $1B+ Lawsuit Against Suno and Anthropic, the handling of rights for training data is also at issue.

Payment Terms and Scope of the $1.5 Billion

Settlement

Payment terms are divided by the form of book publication. For books still in print from traditional publishers, payments are split 50-50 between authors and publishers. For self-published books, authors receive the full amount. Books whose rights have reverted to authors after going out of print are also structured so that authors receive the full amount. The settlement covers about 500,000 titles, totaling $1.5 billion. A flat payment of $3,000 per title has been adopted. By not calculating usage individually for each rights holder, processing has been expedited. On the other hand, the work of confirming rights ownership has become complicated. Determining who the legitimate recipients are is now the focus.

Reports of Excessive Claims by Publishers Mount

This week, multiple authors voiced objections on social media. They contend that publishers are claiming more than their rightful share. Mystery writer April Henry protested a claim by HarperCollins. She said it involved a book whose rights had reverted more than 17 years ago. She also said the publisher had been registered as her employer in her credit information. She explained that no such employment relationship existed.

WTF is HarperCollins playing at? They claimed one of my books on the Anthropic Settlement that reverted back at least 17 years ago AND on the same day I got a credit alert saying they had been added as my employer!

The above is quoted from her post. The fact that a claim on reverted rights and an employer registration occurred on the same day is being questioned. Debate continues over whether this was an isolated clerical error or a systematic claiming policy. Distrust among authors is growing. Questions are also being raised about the payment notification system itself.

Claims by Literary Agencies and Questions

Over Rights Ownership

Victoria Strauss of the writer advocacy blog Writers Beware has compiled complaints from authors. She says the complaints fall broadly into two categories. The first is claims by publishers on works for which their rights have expired. The second is cases where 100% is claimed despite the rightful share being 50%. Both types have been reported by multiple authors with identical details. Strauss said the number of reports in the past two days has been unusually high.

Claims by literary agencies have also been confirmed. Strauss reports that several agencies have filed claims for a share of the proceeds. Since agencies are not rights holders in books, the basis for their claims is unclear. The role of agencies lies in mediating between authors and publishers and managing contracts. The involvement of non-rights-holders in the claims process is causing confusion. Interpretation of the contractual scope between authors and agencies has also become an issue.

Poor Record-Keeping and Procedural Complexity

Strauss said she suspects poor record-keeping rather than assuming bad faith. She expressed caution about attributing to malice what can be explained by poor record-keeping. Some publishers have acknowledged errors and said they have asked Anthropic for corrections. Mary Rasenberger, CEO of the Authors Guild, expressed a similar view. In an interview with The New York Times, she said this was not a deliberate grab by publishers. She explained that she does not consider it an intentional act to harm authors. She indicated her understanding that the cause was poor record management and procedural confusion.

Still, Strauss does not regard the situation as minor. She likened the current complaints to a view through a small crack in a huge wall. She emphasized that the same errors are being reported repeatedly. She said this differs in nature from the sporadic glitches expected in normal large-scale processing. She suggested the possibility of a widespread systemic problem. The scale of about 500,000 titles is making verification difficult. In many cases, histories of out-of-print status and rights reversions appear not to have been accurately tracked by publishers.

The Broader Conflict Over AI Training and

Copyright

This settlement is an important case concerning the boundary between AI training and copyright. It is based on a judicial ruling that separated the legality of training from the illegality of acquiring pirated copies. The flat-payment approach served as a means to avoid prolonged damage calculations. It can be evaluated as a framework that achieved early compensation for rights holders. On the other hand, the work of identifying recipients is more complex than initially expected. Changes to publishing contracts and out-of-print procedures have accumulated over decades. Fragmented rights information is undermining the accuracy of distribution. The confrontation between AI developers and rights holders has moved to the stage of distribution practice. Designing the process from determining compensation amounts to actual payment is the next challenge.

This will also affect rights management systems in the publishing industry. Demand is growing to electronically track contract histories between authors and publishers. Contract reviews to clarify the scope of agency involvement may also proceed. Scrutiny of procurement methods for AI training data will also intensify. There are calls to establish systems to ensure proper licensing channels. The current confusion can be seen as a case that has exposed the fragility of the rights-information infrastructure.

Editorial Opinion

In the short term, congestion in claim-correction procedures is likely to continue. Record checks by publishers will proceed, increasing Anthropic’s review burden. Individual negotiations between authors and publishers will increase, and standards for agency involvement will be clarified.

In the long term, provisions for distributing AI-derived revenue in publishing contracts are likely to become standardized. Rights-ownership databases will be developed, and rights management by publishers and authors will be reviewed. This will also affect the design of settlements in copyright lawsuits against AI developers.

The remaining question is who will accurately manage information on works after rights have expired. Was the framework that separated liability for pirated acquisition from the legality of training appropriate? The question is whether the industry can autonomously build mechanisms to ensure transparency in distribution.

References

• “Authors push back as publishers and agents make claims on Anthropic settlement”, by Anthony Ha — TechCrunch AI, 2026-09-06T20:47:37.000Z (ARR)
• Source URL: https://techcrunch.com/2026/09/06/authors-push-back-as-publishers-and-agents-seek-share-of-anthropic-settlement/