Reservoir Media Sues Suno and Anthropic for $338 Million Over Song Training

Music rights company Reservoir Media Management filed two lawsuits on Oct. 21 seeking a combined $338 million from AI companies Suno and Anthropic over allegations they trained their systems on copyrighted songs and lyrics by artists including Madonna, Johnny Cash and Joni Mitchell.

The damages claims break down to $186.4 million against Suno and $151.4 million against Anthropic, according to the filings, first reported by Billboard.

What does Reservoir allege Suno did?

In the first suit, Reservoir accuses Suno of scraping 1,050 copyrighted songs. The cited works include Johnny Cash's "Ring of Fire," Joni Mitchell's "Both Sides Now" and Madonna's "Papa Don't Preach." Suno operates a platform where users can generate complete AI songs from a single text prompt.

The complaint argues Suno's business was never genuinely about helping people create music. Reservoir states in the filing that Suno "was instead always meant to replace the artists whose works were being unlawfully copied by directly reaching those artists' listeners with AI-generated substitutes which copy the originals."

That claim goes to the core of the legal and market stakes. If a court accepts that Suno built a substitute product on scraped catalogs, the case could reshape how AI music companies obtain training data — and how much they pay for it.

What is Anthropic accused of?

The second suit targets Anthropic, accusing the company of using lyrics from 853 copyrighted songs to train its Claude family of AI models. Works named include Joni Mitchell's "Big Yellow Taxi," Billy Strayhorn's "Lush Life" and Evanescence's "Bring Me to Life."

A representative for Anthropic told Billboard: "We disagree with the allegations and intend to defend this case vigorously." Representatives for Suno and Anthropic did not immediately respond to additional requests for comment.

Both lawsuits accuse the companies of copyright infringement, circumventing access restrictions such as logins and paywalls, and stripping song credits of title, writer and publisher data.

Reservoir also accuses Anthropic of using a text-extraction tool called Newspaper, which removes copyright notices from written material. According to the complaint, Claude is partly trained on lyrics collected from Common Crawl, a web crawler that inadvertently picks up lyrics from licensed sites such as Musixmatch and LyricFind during its sweeps of the open web.

What did Suno already admit in court?

Reservoir's filing cites a statement Suno made in court during a 2024 suit brought by Universal Music Group, Sony Music Entertainment and Warner Music Group. In that earlier case, Reservoir says, Suno admitted to copying "essentially all music files of reasonable quality that are accessible on the open internet."

A separate incident this summer added pressure. A hacker broke into Suno's code and released material that allegedly showed the company had scraped music and lyrics from platforms including YouTube Music, Deezer, Genius and stock music libraries. The exposed data reportedly contained:

• More than 2 million clips in a folder labeled "youtube_music"
• Thousands of hours of music from the lyrics database Genius
• Thousands of hours of music from the streaming service Deezer
• Material from the stock music site Pond5

Suno already faces several copyright infringement lawsuits, including a second case from Sony Music and Universal Music Group.

How much legal exposure does Anthropic carry?

The Reservoir suit is not an isolated event for Anthropic. Last week, another lawsuit accused the company of violating copyright by using 274 Elvis Presley songs to train Claude.

Anthropic has now been sued more than a dozen times for copyright infringement. The most notable case to date was a $1.5 billion settlement over allegations that it trained its AI on millions of pirated books. That figure gives context for the current $151.4 million demand: it represents roughly a tenth of what Anthropic has already paid to resolve one training-data dispute.

Why are these suits stacking up?

Copyright infringement of news articles, music, books and art has become a central legal issue for the AI industry. Multiple AI companies face accusations of ingesting enormous volumes of data to train models including Claude, OpenAI's GPT, Google's Gemini and Elon Musk's Grok.

The scale is measurable. The AI Lawsuit Tracker, a dedicated website, counts 145 copyright infringement lawsuits — more than half of all cases brought against AI companies.

The underlying technology explains the stakes. AI models are trained when companies feed vast amounts of data — text, images, music, videos and software code — into their systems. Combined with synthetic data generated by computers, models learn to produce words, images and video that can appear human-created. Every copyrighted work ingested during that process is a potential claim.

What does Reservoir's CEO say?

Reservoir founder and CEO Golnar Khosrowshahi told Music Business Worldwide that the AI music industry cannot be allowed to exploit songwriters' work.

"When we believe their work has been exploited without authorization, we will hold AI platforms legally accountable to ensure those rights are upheld," Khosrowshahi said. "We will continue to advocate for a framework that supports technological progress while protecting the creative work at the heart of our industry."

That statement signals a dual strategy: litigation to establish liability, plus lobbying for a licensing framework that would make payments to rights holders a structural requirement rather than a courtroom outcome.

What comes next?

The two Oct. 21 filings add Suno and Anthropic to a docket that now includes the majority of all litigation against AI companies. Anthropic has pledged a vigorous defense; Suno's response to the earlier major-label suits suggests it will litigate the scope of fair use rather than settle quickly. The outcomes will help determine whether AI training on copyrighted music requires licenses up front — or damages paid after the fact.