Anthropic and Suno Sued Over 500 Copyrighted Songs Used to Train AI

Independent music publisher Round Hill has sued both Anthropic and Suno in a California court, accusing each company of using its copyrighted song lyrics to train AI systems without authorization, according to Reuters. Round Hill claims Anthropic used lyrics from at least 500 songs to train Claude, while Suno allegedly used the same material for its AI music generator. The publisher is seeking damages that could exceed $1 billion, and warns the total number of affected works may eventually surpass 10,000.

What happened

Detail	Fact
Plaintiff	Round Hill, independent music publisher
Defendants	Anthropic and Suno (separate suits)
Court	California
Songs named so far	At least 500
Potential total works	More than 10,000
Damages sought	Could “approach or conceivably exceed $1 billion”
Artists cited	James Brown, The Kinks, Goo Goo Dolls

Round Hill filed the two suits separately. In the case against Anthropic, it alleges the company pulled lyrics from at least 500 songs it controls in order to teach Claude how to respond to user prompts. The same catalog is at the center of the Suno complaint, where Round Hill claims the AI music-generation startup used those lyrics as part of its own training pipeline.

Round Hill CEO Josh Gruss made the publisher’s position clear in a public statement: “We intend to take these cases to trial and to hold these companies accountable, and we will not accept a resolution that leaves songwriters and artists deprived of their rightful share of compensation.”

Is this Anthropic’s first copyright lawsuit over song lyrics?

No. Anthropic has previously faced legal action from Universal Music Group and BMG Rights Management over separate allegations that thousands of song lyrics were used without permission to train Claude. The Round Hill case adds another plaintiff and a specific new claim tied to at least 500 identifiable works.

Suno is also no stranger to litigation. Universal Music Group and Sony Music have each filed suits against the company over its AI training practices, making Round Hill the latest in a growing line of rights holders pursuing the startup.

These cases sit inside a broader legal debate about whether scraping copyrighted text and audio to train AI constitutes fair use or infringement. No US court has issued a definitive ruling on that question yet, which is part of why Round Hill says it wants a verdict rather than a settlement. For more on how AI companies are navigating safety and legal pressure, see our coverage of OpenAI quietly disbanding its AI safety preparedness team.

Why it matters

For anyone building products on top of AI models, the legal status of training data is a real business risk. If courts side with rights holders, AI companies could face retroactive liability, forced retraining on cleaner datasets, or licensing costs that reshape their pricing. Those costs tend to flow downstream to API customers and businesses that depend on those tools.

The $1 billion damages figure is an estimate from Round Hill, not a court ruling, but it signals the scale of exposure plaintiffs believe is plausible under copyright law. Statutory damages per work can be significant, and with a potential universe of 10,000 songs, the math gets uncomfortable quickly for the defendants.

Businesses that use AI integrations built on Claude or other models with uncertain training provenance should at least be aware that the legal landscape is still being written. That is not a reason to stop using these tools, but it is a reason to watch how indemnification terms evolve in your vendor contracts.

Our take

Round Hill’s decision to pursue trial rather than settle is worth noting. Publishers that accept quiet settlements give defendants a way to avoid precedent-setting rulings, and Gruss is clearly betting that a court victory would be worth more than a check. Whether that gamble pays off depends heavily on how judges interpret fair use in an AI context, something that remains genuinely unsettled.

From a practical standpoint, Anthropic is a well-funded company with experienced legal counsel. Suno is smaller and faces suits from multiple major labels simultaneously, which puts it under heavier financial pressure. Watch whether Suno settles while Anthropic fights, or vice versa. The first company to settle sets a price; the first to win in court sets a precedent.

If you are evaluating AI tools for your business and want a clearer picture of what the options and risks look like, talk to the Lumien team before committing to a platform.

What to do about it

1. Review the terms of service and indemnification clauses in any AI platform contract you currently use, specifically around training data liability.
2. Ask your AI vendor directly whether it has licensing agreements with content rights holders, or relies on fair use arguments.
3. Monitor how these cases progress. A ruling or major settlement in either the Anthropic or Suno cases will likely shift how vendors price and structure their agreements.
4. If your product generates content (text, lyrics, music) using AI, document your due diligence on the tools you use in case you are ever asked to demonstrate reasonable care.

The clearest takeaway: training data provenance is now a vendor evaluation criterion, not just a legal curiosity.

Source: Bing News · Claude AI