Reservoir Wants $340M From Suno, Anthropic Over 1,903 Songs
Ring of Fire is now evidence. A music publisher you have probably never heard of has walked into a federal courthouse in California and accused two of the best-funded AI companies on earth of stealing songs by Johnny Cash, Madonna and Joni Mitchell — and it wants nearly $340 million for them.
Reservoir Media filed the two lawsuits on Wednesday, October 21, one against Suno and one against Anthropic. The details were first reported by Billboard and confirmed when Music Business Worldwide obtained the complaints. If you thought the AI copyright wars were cooling down, think again: this is the industry's litigation blitz entering a new, more aggressive phase.
And this one has teeth. Not just because of the money — though $340 million is the sort of number that makes board meetings go quiet — but because of what Reservoir is asking the court to destroy: every unlicensed copy, "including copies retained in training datasets, internal repositories, model weights, and server infrastructure." That last phrase is the loaded gun. Reservoir isn't just asking for a payout. It wants the model weights themselves wiped.
Reservoir's $340 million two-front war: Suno and Anthropic
The facts, as pleaded. Both cases were filed in the US District Court for the Northern District of California. The plaintiffs are Reservoir Media Management, Inc., its UK subsidiary, and Shapiro, Bernstein & Co. Limited — the historic publishing house wholly owned by Reservoir's UK entity. Reservoir's parent, Reservoir Media, Inc., trades on the Nasdaq, and the complaint describes it as "the first female-founded and -led publicly traded independent music company in the United States."
Reservoir is represented by Richard S. Busch of Adams & Reese — the lawyer who won the 'Blurred Lines' case for Marvin Gaye's family — with Skaggs Faucette LLP as local counsel. This is not a firm that sues for the headlines and settles for the nuisance value.
The Suno complaint names 1,050 compositions whose US copyrights Reservoir says it owns outright or exclusively controls. At the statutory maximum of $150,000 per wilfully infringed work, that is $157.5 million. Two further counts under the Digital Millennium Copyright Act — one for allegedly circumventing access controls ($2.625 million), one for allegedly stripping copyright management information from the works ($26.25 million) — take the Suno total to just over $186 million.
The Anthropic complaint covers 853 works, with the copyright count at $127.95 million plus $2.1325 million and $21.325 million on the two DMCA counts — just over $151.4 million. Combined, the two suits seek roughly $337.8 million: the "nearly $340 million" figure everyone is quoting. We re-ran Reservoir's pleaded damages arithmetic ourselves — 1,050 × $150,000 + $2.625M + $26.25M, and 853 × $150,000 + $2.1325M + $21.325M — and both totals check out exactly.
The songs are the ones your parents know. Joni Mitchell's 'Both Sides Now' and 'Big Yellow Taxi', Johnny Cash's 'Ring of Fire', Madonna's 'Papa Don't Preach', Evanescence's 'Bring Me to Life', plus tracks by Glenn Miller, Kings of Leon and Miami Sound Machine's 'Conga'. The Anthropic complaint adds Billy Strayhorn's 'Lush Life' and Miles Davis's 'So What'.
What Reservoir says Suno did
The complaint's central charge is blunt: Suno "built its entire business model based on copyright infringement at a scale never seen before." Busch goes further, arguing that Suno "made illicit copies of millions of copyrighted musical works, including Reservoir's Works, and then — without any remuneration to the rights holders — created a platform where anyone capable of paying $10 per month can create complete songs based entirely on these illicit copies."
The filings get specific about the plumbing. They lean on a July report by 404 Media, which examined Suno source code obtained by a hacker, and allege Suno used proxy services from Bright Data to scrape "millions of songs and lyrics" from YouTube Music, Deezer and Genius. The numbers quoted are staggering: one YouTube Music workflow saved 2,057,486 artist records; a Genius workflow logged 6,980,653 successful song-page processing records. Reservoir claims several of its own works appear in the leaked code — 'Both Sides Now', 'Ring of Fire', Ben Harper's 'Diamonds on the Inside' — and alleges Suno trained on cover versions alongside originals, "put[ting] a cover recording followed by the 'main' (original) recording of the same song when creating a training data instance." Suno told 404 Media the hacked material was "outdated source code that is no longer in use."
Reservoir also attacks Suno's guardrails, arguing they are trivially evaded. The complaint says Suno won't accept a prompt naming Reba McEntire but will accept one for a female-vocal country song "in the style of Reba McIntyre [sic]" — yes, Reservoir's own lawyers misspelled her surname in the filing; nobody's perfect. It adds that with v6, Suno accepts artist names and merely "redirects" them "toward descriptive musical characteristics," and claims a prompt for "a Miles Davis like Blues track" produced two tracks "clearly inspired by and... unlicensed derivative works of Miles Davis music."
Perhaps the most damaging passage is the one about forgiveness: "Suno's business model was admittedly and clearly: infringe now, make billions by being the first mover, and try to settle on the cheap later. Seeking forgiveness is clearly more profitable than permission, which they obviously could have done but consciously chose not to do."
What Reservoir accuses Anthropic of
The Anthropic suit mirrors the structure: Reservoir alleges the company copied the lyrics of its 853 compositions to train Claude without permission, and that Claude "generates copies of, and derivative works based on, those lyrics" in response to user prompts — output that, the complaint says, nearly replicates some songs in full. "Claude has created infringing output for all 853 of the Works, and likely every song in Reservoir's catalog," the filing claims.
The line written for the press reads: "Anthropic's business model was admittedly and clearly: infringe now, and hope for a two-trillion-dollar IPO later."
Anthropic has pushed back. "We disagree with the allegations and intend to defend this case vigorously," a company spokesperson told Fox News Digital. Suno, for its part, declined to comment.
The DMCA angle is the sharp edge
Don't skim past the Digital Millennium Copyright Act counts. They are the part of these suits that could hurt most. Reservoir isn't just claiming infringement; it claims Suno and Anthropic (a) circumvented technological access controls to get the music, and (b) removed copyright management information — the author and owner details that tell a downstream user whose song it is. DMCA violations carry their own statutory damages and, critically, they sidestep the fair-use defence. You can argue training is fair use all day; it is much harder to argue that stripping attribution metadata is.
This is the same "how you got the data" theory that has been spreading across AI litigation since the Bartz v. Anthropic ruling. Reservoir's complaint quotes Judge William Alsup's June 2025 line directly: "There is no carveout, however, from the Copyright Act for AI companies." And on fair use, the complaint is dismissive: "Building and maintaining a permanent stockpile of copyrighted material for open-ended future exploitation fails every factor of the fair-use analysis." It cites both Alsup's ruling — that keeping a permanent library of pirated copies is not fair use — and the Third Circuit's September 29 decision in Thomson Reuters v. ROSS Intelligence, the first US appeals-court ruling to reject an AI fair-use defence.
Suno is in the more precarious spot
Reservoir's suits land in the middle of an already crowded battlefield. Suno has been the industry's favourite defendant: Warner Music Group settled in November 2025, swapping litigation for a licensing partnership (and selling Suno the Songkick concert platform along the way); BMG struck a global licensing deal in August 2026; Believe followed. Suno's v6 models, launched in September, are built on those licensed foundations.
But Universal Music Group and Sony Music Entertainment never signed — and on September 18 they filed a second lawsuit covering 60,202 recordings. Reservoir's complaint turns Suno's settlements against it: by striking those deals, it argues, Suno "has all but admitted that there is a functioning market for licensing music and sound recordings for training data." In other words: if you can pay Warner and BMG now, you could have paid everyone from the start. The Round Hill Music suit (filed August 17 by the same Busch team) and the Gerencia 360 suit (August 31) add to the pile, along with a first-instance loss to Germany's GEMA at the Munich Regional Court on July 31.
For Anthropic, this is the latest in a cascade of music cases. Concord, Universal Music Publishing and ABKCO sued in October 2023 over roughly 500 songs — and their summary-judgment arguments on whether Claude's training is fair use were heard by Judge Eumi Lee in San Jose on, of all days, the very same October 21, with no ruling yet. A second publisher suit in January 2026 covers more than 20,000 works and seeks over $3 billion; BMG sued in March; Sony Music Publishing and Warner Chappell filed on August 28; Round Hill (again, Busch's team) filed against Anthropic on August 17. Add the $1.5 billion Bartz book-settlement — final approval came on July 20 — and Anthropic's legal department is not getting much sleep.
The common thread across all of it is acquisition: courts and plaintiffs are increasingly focused not on whether AI training can be fair use, but on how the data was obtained. That is a much narrower fight, and a much harder one for the AI companies to win.
The litigation blitz is working — just not in court
Here is my honest position: the music industry is not winning these cases at trial. It hasn't won one. The Bartz case settled for $1.5 billion, the ROSS decision concerned a non-generative retrieval tool, and the fair-use question for generative training is still, technically, unresolved. If you measure "working" by verdicts, the blitz is a failure so far.
But measure it by economics and it is succeeding spectacularly. Suno — the alleged infringer-in-chief — now pays Warner, BMG and Believe, launched licensed v6 models, and raised more than $400 million in June at a $5.4 billion valuation. The lawsuits created the licensing market they claimed didn't exist, then forced Suno to participate in it. Reservoir's complaint essentially says so itself: the settlement money is the proof the infringement mattered.
That's the real dynamic now. The suits are not really about the $340 million — nobody believes these get paid in full — and they're not really about winning a fair-use verdict. They're about price discovery. Each lawsuit is a bid in an auction to decide what a song is worth to an AI company. And with every settlement, the industry's leverage ratchets up: Warner's deal becomes the floor for the next negotiation, and the next, and the next. Botched and shoddy as the rollout of AI music licensing has been — settlements with no transparency, opt-in terms nobody audits — the direction of travel is unmistakable. The question was never whether AI companies would pay for music. It was how much, and who gets to set the price. Lawsuits like Reservoir's are the price-setting mechanism.
That said, the destruction-of-weights demand should not be dismissed as theatre. If any court grants it — impoundment of "model weights, and server infrastructure" — the precedent would detonate across every pending AI case in America. That is the ask to watch.
Update, Oct. 23: the Bright Data fight just narrowed — and the evidence trail keeps growing
There was already a significant development in the wider Suno litigation this week — and it happened in Busch's other cases. On Monday, October 19, the Round Hill Music (filed August 17) and Gerencia 360 (filed August 31) complaints against Suno — both also brought by Busch — voluntarily dismissed their claims against data-scraping firm Bright Data, without prejudice, and filed amended complaints against Suno alone (Music Business Worldwide, October 22). Reservoir's own complaint names only Suno too, but alleges Suno used Bright Data's "services to scrape and incorporate into its training data files like 'youtube_music' which includes over two million music clips of approximately 113,879 hours — the equivalent of thirteen years — of music files scraped from YouTube" (Digital Music News, October 22). Separately, the Anthropic side of Reservoir's case leans on a dataset-search tool run by The Atlantic — the AI Watchdog — which lets anyone check whether specific works appear in known AI training collections (Digital Music News, October 22). Translation: the pleaded evidence is no longer "trust us, they trained on it" — it is searchable.
Frequently asked questions
Who is Reservoir Media?
An independent music publisher whose parent, Reservoir Media, Inc., is listed on the Nasdaq. The complaint calls it "the first female-founded and -led publicly traded independent music company in the United States." It was founded by Golnar Khosrowshahi, who remains CEO. Its catalogue includes publishing rights to songs by Joni Mitchell, Johnny Cash, Madonna and Evanescence, among others.
How much money is Reservoir actually seeking?
Just over $186 million from Suno (1,050 works at up to $150,000 each, plus DMCA damages of $2.625 million and $26.25 million) and just over $151.4 million from Anthropic (853 works, plus $2.1325 million and $21.325 million on the DMCA counts) — roughly $337.8 million combined, or "nearly $340 million." These are pleaded maximums; any actual award or settlement would likely be lower.
When were the lawsuits filed, and where?
Both were filed on Wednesday, October 21, 2026, in the US District Court for the Northern District of California. The story was first reported by Billboard on October 22.
What did Suno and Anthropic say?
An Anthropic spokesperson said: "We disagree with the allegations and intend to defend this case vigorously." Suno declined to comment. Reservoir CEO Golnar Khosrowshahi said the company's responsibility is "to protect the rights and creative legacies of our songwriters, artists, estates, and partners, and when we believe their work has been exploited without authorization, we will hold AI platforms legally accountable."
Has Suno been sued before?
Extensively. The major labels sued in June 2024; Warner settled in November 2025, BMG in August 2026, and Believe reached a deal too. Universal and Sony are still litigating and filed a second suit in September 2026 covering 60,202 recordings. Round Hill Music and Gerencia 360 have also sued, and Germany's GEMA beat Suno at the Munich Regional Court in July.
Does this mean Suno or Anthropic will have to delete their AI models?
Not unless a court orders it. Reservoir is asking for impoundment or destruction of unauthorised copies, "including copies retained in training datasets, internal repositories, model weights, and server infrastructure" — which would effectively mean dismantling the trained models. Courts grant such relief rarely, and typically only after liability is proven. But if granted anywhere, the precedent would reshape every AI copyright case in the country.
Why did Reservoir sue Suno and Anthropic over the same songs?
Reservoir says both companies copied the same catalog — songs like "Ring of Fire," "Big Yellow Taxi" and "Papa Don't Preach" — without licences, but for different purposes: Suno to train a music generator anyone can use, Anthropic to train general-purpose models including Claude. Filing separately in the same court lets the publisher pursue the full $150,000-per-work statutory maximum against each company independently — about $186 million from Suno and $151 million from Anthropic on the copyright claims alone, before the DMCA penalties.
Sources: Music Business Worldwide (which obtained both complaints), Billboard (first report, via syndication), Digital Music News, AI Parabellum, Fox News Digital (Anthropic statement), 404 Media's July report on Suno's hacked source code, Beyond Copying (coverage of the October 21 summary-judgment hearing in the Concord/UMP/ABKCO case), and TechPulse USA's own prior coverage of the ROSS fair-use ruling, Microsoft's internal AI-scraping memos and USA Today's suit against OpenAI.
By Daniel Rao
Daniel Rao covers the collision of AI, copyright and culture for TechPulse USA. He has reported on the AI training-data lawsuits since the first label complaints landed in 2024.
Comments
Post a Comment