The Presley estate lawsuit against Anthropic tests copyright boundaries as estates challenge AI training without permission. The legal clash over Elvis Presley’s catalogue signals a new era where artist heirs confront the tech giants building the next generation of generative tools.
The Elvis Lawsuit and the AI Training Frontier
Authentic Brands Group, the entity that owns both Presley’s music and likeness, filed the suit in federal court on Friday, accusing Anthropic of using 274 Elvis tracks to train its Claude models without a licence. Lead attorney Richard Busch wrote in the complaint that even after Anthropic installed guardrails to limit infringing outputs, the company knew those safeguards were ineffective. The lawsuit claims that from its inception, Anthropic’s AI model, Claude, has created infringing output of copyrighted material from virtually every catalogue on the internet.
How AI Models Learn From Music
AI music platforms work by analyzing large amounts of recorded music to learn sounds, structures, and expressions. The process involves feeding thousands of songs into a model, which then identifies patterns in melody, harmony, and rhythm. Once the model has absorbed those patterns, it can generate new audio based on text or speech prompts. In this case, the plaintiffs argue that Anthropic used Presley’s recordings as training data without consent, turning classic hits into fuel for a commercial product.
The Numbers Behind the Claim
The suit targets 274 specific Presley songs, including Can’t Help Falling in Love, All Shook Up, and A Little Less Conversation. For each allegedly infringed track, the plaintiffs are seeking $150,000 in statutory damages, a figure that would total approximately $41 million (about £31 million) if applied across the entire set.
What This Means for Artist Catalogues
This case represents one of the first major legal challenges brought directly by an artist’s estate against an AI developer. Previous suits have come from record labels and industry groups, but this one targets the estate’s control over a specific body of work. A ruling in favor of the Presley estate could force AI companies to overhaul how they source training data, potentially requiring licences for every song used in model development. Conversely, a victory for Anthropic could affirm that training on publicly available recordings falls under fair use, opening the door for broader AI music generation. The outcome will likely shape how future AI models are built and how artist estates negotiate rights in the years ahead.
The Human Cost of AI-Generated Music
Beyond the legal statutes and financial figures, the Presley estate lawsuit highlights a growing concern among working musicians regarding the devaluation of human artistry. The filing notes that Anthropic’s model, Claude, has generated infringing outputs spanning virtually every catalogue on the internet, raising alarms about the saturation of AI “slop” in the market. Last week, Spotify confirmed it had removed approximately 50 million songs flagged as AI-generated from its recommendations over the past year, a figure that underscores the scale of automated content flooding streaming services. Musicians have repeatedly voiced fears that their voices and compositions are being harvested without consent to train systems that could ultimately replace them, transforming the creative ecosystem into one where human-made music competes directly with machine-generated approximations.
This tension is not merely theoretical; it affects the livelihoods of the artists whose work fuels the technology. The lawsuit’s emphasis on the 274 Presley tracks used without a licence illustrates a broader industry pattern where AI developers ingest copyrighted recordings to refine their models. With the estate seeking $150,000 in statutory damages per track—potentially totaling $41 million—the case underscores the financial stakes of unauthorized training. As the legal battle progresses, the music community watches closely, recognizing that the outcome could determine whether future AI systems are built on licensed archives or on the uncompensated labour of the artists they mimic.
The Ethics of AI Music Training
Anthropic’s approach to training its Claude models has drawn sharp criticism regarding the origins of its data. According to the lawsuit, the company utilized a project internally described as an “effort to destructively scan all the books in the world,” which previously involved downloading pirated books before settling a 2025 copyright infringement lawsuit from authors for $1.5 billion. The suit alleges that Anthropic knowingly used Presley’s recordings as training data despite being aware that safeguards were ineffective at preventing copyrighted output. This methodology reflects a broader industry pattern where AI developers ingest vast quantities of copyrighted material to refine their systems, raising fundamental questions about whether such use constitutes fair use or systematic infringement.
Musicians and rights holders have voiced escalating concerns that their work is being harvested without consent to train commercial AI systems. The lawsuit notes that Anthropic’s model has generated infringing outputs spanning virtually every catalogue on the internet, contributing to the saturation of AI “slop” in the market. Last week, Spotify confirmed it had removed approximately 50 million songs flagged as AI-generated from its recommendations over the past year. Working musicians fear that their voices and compositions are being used without permission to train systems that could ultimately replace them, transforming the creative ecosystem into one where human-made music competes directly with machine-generated approximations. The Presley estate case underscores the financial and ethical stakes of unauthorized training, with the plaintiffs seeking $150,000 in statutory damages per track to address the devaluation of human artistry.
Sources
- Anthropic sued for allegedly using 274 Elvis Presley songs to train AI models
- Anthropic
- How Claude designed protein binders against scorpion neurotoxins
Photo: TechCrunch (CC BY 2.0)
