Sony Music, Warner sue Anthropic, alleging a 'brazen campaign' of intellectual property theft

The core allegations are severe. The publishers accuse Anthropic of 'blatant theft' in using thousands of copyrighted works — including books, lyrics, and sheet music — to train Claude. More notably, the new case escalates beyond existing copyright suits by specifically alleging that Anthropic obtained millions of copies of books through 'illegal torrenting,' including works containing lyrics and sheet music. This means plaintiffs are not just alleging 'scraping' of publicly available web content, but that Anthropic used more invasive means like P2P torrenting to acquire training data — an uncommon charge in previous AI copyright litigation.

Anthropic's response was brief and firm. A spokesperson wrote in an emailed statement to TechCrunch: 'We disagree with the publishers' claims and we intend to defend ourselves robustly in court.' This is far from the first time Anthropic has faced copyright litigation over training data. In fact, some of the same lawyers behind this new lawsuit also represent Concord Music Group and Universal Music Group in a case filed in January, and they led the landmark Bartz v. Anthropic case — in which a group of authors accused Anthropic of using copyrighted works to train products like Claude, and Anthropic was ordered to pay $1.5 billion in damages. In other words, the music and publishing industries have developed an organized, systematic litigation strategy against Anthropic.

While the cases share core arguments, the key differences lie in scope and the nature of the evidence. The latest lawsuit is particularly broad, building on prior cases while introducing 'illegal torrenting' as an explicit allegation for the first time — torrenting typically involves obtaining files from P2P networks, making the infringement more clear-cut than mere scraping of public web pages, with a more direct evidentiary path. The timing is especially delicate for Anthropic: just days earlier (August 28), the company made headlines with its 'automated alignment researcher' paper showing AI systems can autonomously improve training; now it must simultaneously face dual copyright pressure from both the music and publishing industries. While AI labs sprint at the technological frontier, the 'historical debts' of training data compliance are being settled one by one.

The significance of this lawsuit for the entire AI industry far exceeds the fate of one company. First, it shows that rights holders are shifting from 'case-by-case protests' to 'industry-wide settlements' over AI training data — music publishers, book authors, news organizations, and photo libraries are joining forces, putting unprecedented judicial pressure on AI companies' 'fair use' defenses. Second, if the 'torrenting' allegation holds, it would directly undermine the defense that 'AI companies only scrape publicly available data' — if training data indeed came from P2P networks, whether the data was legally obtained ceases to be a gray area. Third, the precedent of massive damages (the $1.5 billion Bartz ruling) is changing AI companies' risk calculus — training data compliance is no longer just a legal department issue but a core risk factor directly tied to company valuation. Going forward, Anthropic's defense strategy and whether the case proceeds to substantive trial will become benchmark cases in the AI copyright battle.

For AI industry observers, the most notable aspect of this lawsuit is not the legal details themselves but the trend it reveals: the era of AI companies 'scraping everything' and growing wildly is coming to an end. As music publishers, book authors, news media, and even photo libraries take up legal arms, the compliance cost of AI training data is becoming a new barrier to entry. For smaller AI companies, this means the 'train first, apologize later' path is increasingly dangerous; for the open-source community, the transparency and traceability of training data are becoming unavoidable topics. Technology always outpaces law, but the law always catches up eventually — this case is a step in that catching-up process.