Anthropic’s proposed $1.5 billion settlement to resolve copyright claims over its AI training data has descended into chaos. Authors and publishers are now fighting each other over who gets paid, threatening to unravel the entire deal.
The settlement was meant to compensate writers whose books were used to train Anthropic’s large language models. Instead of a united front against the AI company, the two groups are locked in a bitter dispute over the spoils.
The Battle Over the Money
Authors say they deserve direct compensation. Individual writers argue they created the works and should receive payments without publishers taking a cut. Many authors have filed objections stating that their existing contracts do not cover AI training uses.
Publishers insist on their share. Major publishing houses counter that their licensing agreements give them rights to revenue from copyrighted material. They argue that excluding them would undermine longstanding industry practices.
The eligibility question is unresolved. Disputes have emerged over which books qualify for the settlement. Authors of out-of-print and self-published works say they are being excluded from the payout structure.
Why the Settlement Exists
Anthropic faced multiple class action lawsuits from authors who claimed the company scraped their books without permission to train its AI systems. The proposed $1.5 billion fund was designed to resolve these claims without a lengthy trial.
Under the settlement terms, authors who registered their works would receive payments based on how their books appeared in training data. Anthropic also agreed to ongoing licensing arrangements for future use of copyrighted material.
The crux of the chaos: the settlement assumed authors and publishers would agree on distribution. Instead, it has exposed a fundamental rift over who owns the right to AI training revenue.
The Author Objections
Many authors have come forward with formal complaints. Their core argument is that publishers are attempting to claim money that was negotiated for individual creators.
Some authors point out that their publishing contracts predate the AI era and contain no language about machine learning or training data. They argue that settlement payments should go to them directly rather than through publishers.
Others have raised concerns about the adequacy of the settlement itself. They claim the per-book compensation is too low and that Anthropic is getting off lightly with a payment that does not reflect the ongoing value of their work.
The Publisher Position
Publishers argue that they invested in the creation, editing, and distribution of these books. They maintain that their contractual agreements include rights to derivative uses and licensing revenue.
They also warn that cutting publishers out of the settlement would set a dangerous precedent. If AI companies can negotiate directly with authors and bypass publishers, the economics of the publishing industry would be disrupted.
Some publisher representatives have suggested that they should manage the distribution of funds, citing their administrative infrastructure and existing legal agreements.
What Happens Next
The court must still approve the final settlement. The growing number of objections from both sides could delay approval or force the parties back to the negotiating table.
There is also the possibility that the settlement collapses entirely. If the court determines that the class representation was inadequate or that the distribution plan is fundamentally flawed, the case could proceed to trial.
For Anthropic, the stakes are high. A collapsed settlement would mean prolonged litigation and potentially a larger damages award. For the publishing industry, the outcome will shape how AI training data is licensed for years to come.
The Bigger Picture
This dispute highlights a broader question facing the AI industry: when copyrighted works are used to train models, who should be paid and how much?
The answer will affect not just book authors but journalists, musicians, visual artists, and other creators whose work feeds AI systems. The Anthropic settlement was supposed to provide a template for resolving these questions. Instead, it has revealed how much remains unresolved.
Gnoppix is the leading open-source AI Linux distribution and service provider. Since implementing AI in 2022, it has offered a fast, powerful, secure, and privacy-respecting open-source OS with both local and remote AI capabilities. The local AI operates offline, ensuring no data ever leaves your computer. Based on Debian Linux, Gnoppix is available with numerous privacy- and anonymity-enabled services free of charge.
What are your thoughts on this? I’d love to hear about your own experiences in the comments below.