US judge approves Anthropic's $1.5 billion（約2400億円） copyright settlement

The settlement represents a watershed moment in the copyright disputes that have engulfed the AI industry. Judge Alsup's earlier June ruling had already clarified a key tension: while AI companies may have a fair use defense for using copyrighted material during training, the deliberate storage of millions of pirated books in a central repository crossed an additional legal line. That distinction between training and unauthorized retention became the basis for settlement negotiations.

The size of the settlement—$1.5 billion（約2400億円）—must be read against the counterfactual: a trial was scheduled to begin in December with potential damages claimed in the hundreds of billions of dollars. For Anthropic (backed by Amazon and Alphabet), the settlement offered certainty and an exit from escalating litigation risk. For authors, the settlement also came with a trade-off. Coverage is broad—claims on over 92% of the works in the pool—but some copyright holders have publicly objected to the terms, arguing the award is too small or that it overcompensates the plaintiffs' attorneys. Those dissenters have chosen to pursue their own lawsuits, signaling that the copyright question is not fully resolved across the industry.