Round Hill Sues Suno, Anthropic — Illegally Scraped Music Isn't Fair Use
Round Hill Music, an independent music publisher with a $1.1 billion portfolio of music copyrights, filed two separate copyright lawsuits on August 17, 2026 . Both allege the same core conduct: that the defendants trained AI models on Round Hill's copyrighted music without a license and without permission.
The Suno suit covers both the compositions and the recordings Round Hill controls. It alleges Suno built its training dataset by scraping copyrighted music directly off the open web, including YouTube, without permission or a license, and used that scraped material to train the AI models behind its song-generation platform. The Anthropic suit is narrower in scope, covering only lyrics and compositions rather than recordings, since Anthropic’s Claude is a text model rather than an audio generator, but it makes the same category of legal claims: that Claude was trained on copyrighted lyrics without a license, and that the underlying material was obtained through the same kind of unauthorized copying and access-control circumvention alleged against Suno. Round Hill's complaint cites an example where Claude, asked to rewrite several of the songs at issue for a modern audience, generated output the company says stayed extremely close to the originals in structure and melodic hooks, changing little beyond surface wording.
Each suit seeks damages that could exceed $1 billion. Round Hill's plaintiffs, the parent company and five affiliated funds, own or control 14,364 musical compositions and 16,873 sound recordings, including songs like Bonnie Tyler's "Total Eclipse of the Heart," the Goo Goo Dolls' "Iris," and James Brown's "I Got You (I Feel Good)." The initial filings name roughly 500 songs, described as a bellwether sample, with Round Hill saying it plans to expand that to 10,000 or more compositions as the cases proceed.
CEO Josh Gruss has said the company will not settle. Its lawyer, Richard Busch, the same attorney who won the "Blurred Lines" case for the Marvin Gaye estate, said it plainly: "there is nothing fair about that use."
Why Round Hill Is Different From Warner, UMG, Sony And BMG
Round Hill isn't the first rights holder to sue Suno, and this isn't even the first time a court has weighed in on Suno's fair use defense. A Munich court ruled against Suno on July 31, 2026 , in a case brought by the German collecting society GEMA, finding that Suno infringed copyright by training on GEMA's repertoire and rejecting Suno's argument that training didn't require a license. That ruling isn't binding on any US court, since it applies German and EU law rather than American fair use doctrine, but it's the first real-world test of Suno's core defense, and Suno lost it. The RIAA, acting on behalf of Universal Music Group, Sony Music, and Warner Music, sued Suno and Udio in June 2024, alleging the companies trained their models on the labels’ recordings without authorization, including through the same stream-ripping method Round Hill now alleges. Those cases are still active for Universal and Sony, consolidated in Boston before Chief Judge F. Dennis Saylor IV, with Suno defending itself by arguing that training on legitimately acquired recordings qualifies as fair use under US law. Warner settled with Suno in November 2025 , exiting the litigation for a licensing deal. BMG, an independent publisher, never sued at all, choosing a direct licensing alliance with Suno instead. The Udio case followed a similar split: Universal settled in October 2025 and Warner settled the same week it settled with Suno, while Sony is still litigating against Udio as well.
Round Hill is betting on a broader legal strategy than any of them, one that doesn't hinge entirely on how the fair use question eventually gets resolved in those other cases.
The Bright Data Mechanism
Round Hill’s complaints make four separate legal arguments. The first is direct copyright infringement under the Copyright Act: both suits allege Suno and Anthropic simply copied Round Hill's compositions and recordings without a license. The second and third come from the DMCA: circumvention of technical access controls, and separately, removal of copyright management information, the embedded data identifying a work's owner and terms of use, which Round Hill alleges both companies stripped out during scraping. The Suno complaint names a third defendant, Bright Data, an Israeli data-scraping company, alleging it supplied the proxy networks and scraping tools Suno used to pull music and lyrics off licensed platforms including YouTube, making Bright Data liable for contributory infringement.
The fourth argument goes directly at fair use itself, rather than around it. Round Hill's lawyers argue that building and maintaining what they call a permanent stockpile of copyrighted material for open-ended future use fails every factor of the fair-use test on its own terms: the purpose is acquisitive rather than transformative, entire works are retained without limitation rather than used and discarded, and the retained copies function as market substitutes that displace the licensing deals Round Hill would otherwise have made. That's a direct challenge to the reasoning AI companies have used to win fair-use arguments elsewhere, not an attempt to route around it through a technicality.
There's real precedent behind that framing. In June 2025, Judge William Alsup ruled in the Anthropic books case that training an AI model on copyrighted text could itself be transformative fair use, while separately ruling that storing millions of pirated books in a permanent library did not share that protection. Round Hill's complaint takes that same distinction, acquisition versus use, and turns it into the center of its case rather than a side argument.
Richard Busch's own language in the filing makes the strategy explicit: AI companies have "waxed poetic about the necessity of progress," he wrote, but there is "no reason, other than rote expediency," for that progress to come at copyright holders' cost. Round Hill's suits also note that Suno's and Anthropic's valuations, built in part on this material, undercut their own argument that paying for it would have been prohibitively expensive, since other AI companies and rights organizations have built training datasets by actually licensing the material.
Winning any one of these four arguments would establish that Suno never had the right to the music in the first place, and it would get Round Hill paid for what was taken. It would not guarantee that payment reaches anyone beyond Round Hill.
Winning Doesn't Mean Anyone Else Gets Paid
The AFM’s suit against Universal Music and Warner Records shows exactly why that gap matters. Those labels already got paid when they settled with Suno and Udio. According to the union's complaint, the session musicians whose recordings actually trained those models have seen none of it.
The American Federation of Musicians, the union representing session players, has a standard contract with the major labels called the Sound Recording Labor Agreement. That contract includes a "new use" clause: when a label licenses a recording for a new commercial purpose beyond its original release, the musicians who played on it are owed additional payment. The AFM argues that Universal's October 2025 settlement with Udio and Warner's November 2025 settlements with both Udio and Suno qualify as exactly that, and that neither label paid out under the clause. The union sued both companies on June 5, 2026, and amended the complaint on July 24. Universal's response, filed in early August, argues the clause "does not set the price. It says where the price lives," meaning in a separate AI-specific rate the union and the labels haven't yet negotiated. In practice, that treats an unset price as equivalent to no obligation at all.
The Money Already Flowing To AI-Generated Acts
While that fight over who gets paid drags on, AI-generated music is already producing real, measurable revenue elsewhere in the same pipeline. A Kapwing analysis published in July 2026 tracked earnings for the most popular AI music acts using streaming and follower data. A synthetic "1950s blues" persona called Enlly Blue has earned an estimated $380,800 from roughly 95 million Spotify streams, despite not being a real performer, just a voice and image generated to sound like one. A YouTube channel using AI-assisted production, Cherry 葵 Nightcore, has pulled in an estimated $2 million from nearly a billion views. A Suno-powered act called Masters of Prophecy earned an estimated $652,000 on YouTube alone, after gaining 30 million subscribers in four months, a growth curve researchers at Garbage Day flagged as statistically implausible for organic reach. Xania Monet, an AI vocal project built by feeding a Mississippi poet's lyrics into Suno, became the first AI act to chart on radio airplay, then signed a multimillion-dollar deal with Hallwood Media.
Irving Azoff, the artist manager who founded the Music Artists Coalition, named the pattern directly after UMG's Udio settlement: "We've seen this before. Everyone talks about partnership, but artists end up on the sidelines with scraps."
What A Jury Verdict Would And Wouldn't Fix
Warner's settlement with Suno ended its litigation without a court ever ruling on the underlying question. BMG avoided that question entirely by never suing in the first place and going straight to a licensing deal. Neither company has disclosed the terms of its agreement with Suno. What is public is the AFM's allegation: that the labels got paid and the session musicians whose recordings trained the models did not. Getting paid and paying the people who made the thing you're getting paid for are two separate steps, and right now only one side of that has a documented outcome, and only for whoever had the size and leverage to force it. Round Hill has that leverage. A synthetic act's listeners, and the session musicians whose recordings trained the model producing what those listeners hear, do not. Whatever a jury decides about Suno, that imbalance doesn't go away. It just moves to whoever's next in line to sue.
Loading article...