Sony And Warner Sue Anthropic As AI Copyright Fight Moves To Music

The AI copyright fight has moved deeper into music, and Anthropic is now facing one of the sharpest claims yet from the entertainment industry.

Sony Music Publishing and Warner Chappell Music have sued Anthropic in a federal court in Northern California, accusing the company of using copyrighted music works to train Claude without permission. Axios reports that the complaint names Anthropic and two of its co-founders, Dario Amodei and Benjamin Mann, while The Verge says the publishers are seeking statutory damages that could reach $150,000 per infringed work.

The allegation is not simply that Claude can discuss music. The publishers argue that Anthropic copied or scraped large volumes of protected material to improve its models, including lyrics and associated copyright information. The complaint also claims that some copyright management information was removed, which could create separate legal exposure if proven.

Anthropic disagrees with the claims and says it intends to defend itself. That is expected. Every major AI company has argued in some form that training on large datasets is either fair use or necessary for machine learning to function. Copyright owners see it differently. Their position is that AI companies built valuable commercial products by ingesting protected works at scale without negotiating licenses.

Music makes the fight more complicated than books or news articles. Songs carry layered rights: compositions, lyrics, recordings, publishing rights and licensing structures. A model that can reproduce or closely imitate protected lyrics raises different questions from a model that summarizes a public article or answers a general question about an artist.

This is also bad timing for Anthropic. The company has been building a public image around safety, enterprise trust and controlled deployment. But it is now dealing with legal pressure on several fronts, including copyright, government procurement and AI safety scrutiny. The Claude brand may be strong, but the business is increasingly operating in the same legal storm as OpenAI, Google and Meta.

For the AI industry, the case adds pressure for licensing deals. If courts decide that training on copyrighted works without permission is not fair use, the cost of building frontier models could rise sharply. AI labs may need to pay publishers, record labels, image libraries, news organizations and other rights holders before training or fine-tuning commercial models.

For creators, the case is about more than money. Songwriters and publishers fear a world where AI models absorb human work, reproduce familiar styles and reduce the market for licensed creative output. AI companies counter that models learn patterns rather than store works like a database, and that strict licensing requirements could slow innovation or entrench only the richest labs.

The practical middle ground may eventually be licensing, transparency and opt-out systems. But that will not happen cleanly without court pressure. This lawsuit is one more sign that the AI economy is now being forced to answer an old question in a new form: who gets paid when machines learn from human culture?

The music industry has fought technology before, from file sharing to streaming. This time, the fight is not only about distribution. It is about training. If the publishers win, AI companies may have to rethink how they source data. If Anthropic wins, the fair-use argument becomes much stronger across the AI industry.