Sony, Warner Music sue Anthropic over songs used in AI training

The plaintiffs, specifically Sony Music Publishing and Warner Chappell Music, two of the "big three" global music publishers that collectively control an immense catalog of musical works, claim that Anthropic "pirated hundreds of song lyrics and sheet music" from iconic artists to fuel its AI development. The complaint specifically cited compositions from legendary figures like The Beatles and Michael Jackson, alongside contemporary superstars such as Taylor Swift, underscoring the vast breadth and commercial significance of the intellectual property allegedly appropriated. These compositions, representing billions of dollars in economic value and cultural heritage, were reportedly used to train Anthropic’s Claude models to respond to human prompts, enabling the AI to generate text, including song lyrics, in a manner that allegedly infringes upon existing copyrights.

The complaint meticulously details how Anthropic purportedly obtained these invaluable creative works. It alleges that the AI company illegally acquired the publishers’ lyrics and sheet music through "torrent downloads" – a method commonly associated with online piracy – to build the foundational knowledge base for Claude. This specific allegation of torrenting is crucial, as it suggests a direct and deliberate act of unlicensed acquisition, potentially undermining any future "fair use" defense that Anthropic might attempt to mount by claiming that the data was merely scraped from publicly available sources without malicious intent. Furthermore, the publishers assert that Claude AI models are not only trained on this illicitly obtained data but can also reproduce copyrighted lyrics "verbatim" when prompted, offering concrete evidence of direct infringement rather than merely transformative use. This capability for direct reproduction is often considered a strong indicator of infringement in copyright law.

Beyond verbatim reproduction, Sony and Warner also contend that Anthropic leveraged their lyrics to teach Claude to "generate vast quantities of purportedly ‘new’ AI-generated song lyrics, which compete with Music Publishers’ legitimate copyrighted works as harmful market substitutes." This claim introduces the critical concept of market harm, a key factor in fair use analysis. The publishers argue that by creating AI-generated content that mimics or is derived from their protected works, Anthropic is directly competing with the very creators whose work it has allegedly exploited, thereby devaluing their intellectual property and depriving them of potential licensing revenue. This scenario paints a picture where AI becomes a direct economic threat to the creative industries it purports to serve, unless proper licensing and compensation frameworks are established.

This lawsuit is the latest, but certainly not the last, in a rapidly expanding wave of legal actions brought against prominent technology companies by a diverse array of copyright owners. Authors, literary publishers, music labels, visual artists, and news organizations have all launched legal challenges, asserting that their intellectual property has been unlawfully used to train powerful generative AI systems without permission or compensation. The core legal question revolves around whether the act of ingesting copyrighted material for the purpose of training an AI model constitutes a "fair use" under copyright law, or whether it is an act of infringement. AI developers often argue for fair use, citing the transformative nature of AI output and the idea that training data is merely an input, not a directly copied output. Copyright holders, conversely, argue that the sheer scale of copying involved and the potential for AI-generated content to compete with original works tips the balance towards infringement.

Anthropic is no stranger to these legal challenges. In 2023, Universal Music Group (UMG), the world’s largest music company and a direct competitor to Sony and Warner, initiated its own lawsuit against Anthropic over the alleged use of copyrighted song lyrics in AI training. That case, which remains ongoing, signals the music industry’s unified front against what it perceives as rampant copyright infringement by AI developers. The current filing by Sony and Warner underscores the industry’s determination to protect its vast catalogs and secure fair compensation in the evolving digital landscape dominated by AI.

Adding another layer of complexity to Anthropic’s legal history, the company became the first major AI firm to settle one of these high-profile disputes last year. It paid an estimated $1.5 billion to resolve a class-action lawsuit brought by a group of authors who alleged that their copyrighted books were used to train Anthropic’s AI without permission. This substantial settlement, while specific to literary works, set a significant precedent and drew sharp criticism from the music publishers in their recent complaint. Sony and Warner explicitly stated, "Anthropic clearly considers that to be just the cost of doing business given that its entire business model continues to be built on copyright theft." They further emphasized the perceived inadequacy of the previous settlement, adding, "And $1.5 billion is obviously not a large enough settlement to deter infringing conduct by a company that has parlayed such mass infringement into a staggering $2-trillion-dollar valuation." This statement highlights the publishers’ belief that Anthropic’s financial success is directly tied to its alleged copyright violations and that current legal remedies are insufficient to curb such practices. It also underscores the massive financial stakes involved, framing the dispute not just as a legal battle but as a contest over the fundamental economics of the digital creative economy.

The legal remedies sought by Sony Music Publishing and Warner Chappell Music are substantial. They are seeking statutory damages of up to $150,000 for each infringed copyright. Under U.S. copyright law, statutory damages provide a powerful tool for copyright holders, allowing them to claim a fixed amount per infringement without needing to prove actual financial losses, which can be difficult to quantify in AI training contexts. Given the claim of "hundreds of song lyrics" being infringed, the potential total damages could quickly escalate into hundreds of millions, if not billions, of dollars, posing a significant financial threat to Anthropic. Furthermore, the labels are demanding a court order, known as an injunction, barring Anthropic from using their works. Such an injunction would have profound implications for Anthropic’s business model, potentially requiring it to retrain its AI models on "clean" data, a process that could be immensely costly, time-consuming, and technically challenging. It would also set a precedent for how AI companies must source and manage their training data moving forward.

Spokespeople for Anthropic, Sony Music, and Warner Music did not immediately respond to requests for comment on Monday, a common practice in the early stages of high-stakes litigation. However, the legal filings themselves provide ample insight into the escalating tension and the divergent views on intellectual property rights in the age of generative AI.

This lawsuit arrives at a critical juncture for the burgeoning AI industry. The legal landscape around AI training data remains largely undefined, with courts and legislatures grappling with how to apply existing copyright laws to novel technological applications. The outcome of cases like this could shape the future development of AI, potentially forcing companies to adopt more robust licensing models, invest in ethically sourced datasets, or face significant financial penalties and operational restrictions. For the music industry, which has historically battled against piracy and fought for fair compensation in various technological shifts – from radio to digital downloads to streaming – this AI wave represents another existential challenge. The industry’s unified stance, exemplified by the actions of UMG, Sony, and Warner, signals a determined effort to ensure that creators and rights holders are not left behind as AI reshapes the creative landscape. The resolution of these cases will undoubtedly set precedents that reverberate across all creative sectors, determining whether AI truly becomes a tool for empowerment and innovation, or a new frontier for exploitation.