In one of the most landmark copyright disputes in the AI industry, the debate over "dividing the cake" has once again drawn attention. According to the latest court documents, the plaintiff's legal team has voluntarily reduced its requested attorney fees from $300 million to $187.5 million in relation to the previous $1.5 billion copyright settlement with AI giant Anthropic.
This move is not an act of generosity by the law firm, but rather a result of pressure from both the court and the defendant. Previously, Anthropic and the judge presiding over the case had opposed the $300 million "sky-high" attorney fee, arguing that the amount was excessively inflated and involved unreasonable cost-sharing plans.
The 1.5 Billion Dollar Settlement: The Cost of AI Training
As the developer of the Claude series of models, Anthropic agreed to pay $1.5 billion last August to settle allegations that it used hundreds of thousands of pirated books to train its AI models. This is not only the largest copyright class-action settlement to date, but also sets a precedent for the industry: Anthropic has pledged to destroy the relevant pirated datasets and ensure that future commercial models do not use such materials.
According to the agreement, each copyright holder of a protected work will receive more than $3,000 in compensation. This high compensation standard is seen as a major victory for traditional content creators against the AI "free feeding" model.
Attorney Fee Dispute: Who Represents the "Collective Interest"?
The reduction in fees mainly focuses on the internal structure of the legal team. The original $300 million request included $75 million allocated to three other law firms not officially designated as representatives of the class action. The judge clearly stated that the lead law firm cannot unilaterally "appoint others" to share responsibility and divide the money. In the newly submitted documents, the attorneys admitted to accepting the court's suggestion, stating that the current $187.5 million request (which accounts for about 12.5% of the settlement fund) will be based solely on the workload of the officially designated lead attorneys.
April Hearing: The Settlement Enters Its Final Stage
The U.S. District Court for the Northern District of California has scheduled a hearing on April 23 to decide whether to finally approve the revised settlement agreement. With this key obstacle resolved, this long-standing copyright case in the AI industry is expected to officially come to an end.
For the entire AI industry, this substantial settlement is not only a payment for past actions, but also like a ticket into the "compliance era." As legal boundaries become clear, the future collaboration model between AI models and copyright holders will gradually shift from "use first, then litigate" to "obtain permission first, then train."
