Music giants demand up to $150K per song from AI startup over alleged training data theft
Sony Music and Warner Chappell have filed a lawsuit against Anthropic, claiming the AI company used a vast number of copyrighted songs to train its Claude model without authorization. The plaintiffs cite examples like 'Eye of the Tiger' and are seeking up to $150,000 in damages for each infringed work. They describe the alleged practice as a deliberate and widespread violation of their intellectual property rights.
The lawsuit marks another front in the escalating conflict between creative industries and generative AI developers. Music publishers argue that training large language models on copyrighted lyrics and compositions without licensing undermines the economic value of their catalogs. The cited damages, up to $150,000 per work, reflect statutory maximums for willful infringement, signaling the plaintiffs’ intent to treat the matter as systematic rather than incidental. As AI models increasingly ingest vast text corpora, rights holders across sectors are testing legal boundaries, with outcomes likely to shape future licensing norms for training data.
This case could influence how AI companies source training material, potentially forcing them to negotiate licenses or build datasets from public-domain or opt-in content. If successful, it may raise costs for model development, affecting startups and researchers with limited budgets. Songwriters and publishers could gain stronger leverage over AI-generated music, while consumers might see changes in how AI tools handle lyrics. The ruling may also set precedent for other creative industries, though its broader societal impact depends on how courts balance innovation incentives with intellectual property protections.