← Back to the feed

Sony Music and Warner sue Anthropic for intellectual property theft

Developed over time first seen 2 months ago

Far Out ·

Sony Music Publishing and Warner Chappell Music sued Anthropic, accusing the AI firm behind Claude of a "brazen campaign" of illegally torrenting, scraping and downloading copyrighted songs to train and profit from its models. The complaint, filed on 28th August, names chief executive Dario Amodei and co-founder Benjamin Mann as individual defendants, and the publishers describe the conduct as "one of the largest and most blatant ongoing thefts of intellectual property in history." Anthropic has rejected the claims, saying it intends to "defend ourselves robustly in court," and no further developments have emerged since the filing.

The publishers are seeking damages of up to $150,000 per work across "tens of thousands" of titles, plus $25,000 for each instance of stripped copyright data, with total damages potentially reaching several billion dollars. Songs cited include Bon Jovi's "Livin' on a Prayer", Marvin Gaye and Tammi Terrell's "Ain't No Mountain High Enough", Leonard Cohen's "Hallelujah" and Taylor Swift's "Paper Rings". The case joins a string of other music-industry suits against Anthropic from Universal Music Group, Concord, ABKCO, BMG and Round Hill Music, and follows a June 2025 ruling that ordered the company to pay $1.5 billion after a court found it had downloaded over seven million pirated books to train its AI models.

  • Sony and Warner Chappell sued Anthropic over alleged mass copyright infringement
  • Damages sought could reach several billion dollars for "tens of thousands" of songs
  • Anthropic denies wrongdoing; case adds to string of prior music and book lawsuits

New here? Start with this

Anthropic, the company behind the AI chatbot Claude, trains its models by feeding them huge amounts of text and other material found online. Music publishers Sony Music Publishing and Warner Chappell Music say that among this material were song lyrics and recordings that Anthropic had no right to use, allegedly obtained through illegal downloading and web scraping rather than proper licensing agreements.

The dispute centres on how AI companies should be allowed to gather the data used to train their systems, and whether copyright holders deserve payment or permission when their work is used this way. Anthropic, led by chief executive Dario Amodei, is a major player in the AI industry, and Claude competes directly with products like ChatGPT. Music publishers control the rights to songwriting and lyrics, separate from the record labels that own recordings, and are increasingly asserting those rights against tech firms.

This case is part of a broader wave of legal action from the entertainment and publishing industries against AI developers, with other music companies and, previously, book publishers bringing similar claims against Anthropic. The outcome could shape how much AI firms must pay, or change, in the way they source training material in future.

Both sides, in good faith

The strongest fair case each way — we don't pick a winner.

The case for

Sony Music Publishing and Warner Chappell argue that songwriters and rights-holders spent decades building catalogues protected by clear copyright law, and that Anthropic obtained many of these works through torrenting and scraping rather than licensing, bypassing the market entirely. They contend that a company built on respecting rules of consent and compensation should not be permitted to profit from wholesale, unauthorised copying simply because the output is transformed by an algorithm, especially given a prior court finding against Anthropic over pirated books. Holding executives personally accountable, in their view, reflects the deliberateness of a decision made at the top of the company rather than an incidental technical oversight.

The case against

Anthropic and its supporters would argue that training AI models on widely available text and data, including song lyrics, plausibly falls within fair use because the process extracts statistical patterns rather than reproducing or distributing the original works themselves. They would point out that copyright law was not designed with machine learning in mind, and that innovation in the field depends on access to large, diverse datasets that no company could feasibly license work-by-work in advance. From this perspective, aggressive litigation and maximalist statutory damages claims risk being used less to compensate specific harm and more to entrench incumbent industry gatekeepers against a disruptive new technology.

More coverage

AI Entertainment Music Technology

Read the full article at the source →