Round Hill Music filed two major copyright lawsuits last week seeking statutory damages of up to $150,000 per work, potentially totaling hundreds of millions or exceeding $1 billion in each case. The music publisher is targeting Anthropic, maker of Claude, and music-generation startup Suno, along with data-scraping firm Bright Data. What sets these filings apart from previous AI copyright cases is their explicit demand that courts impound or destroy the trained AI models themselves, not just the original files used to train them.

What Makes These Lawsuits Different From Past AI Copyright Cases?

The legal strategy in both complaints goes beyond what previous settlements have required. When Anthropic settled with book authors for $1.5 billion in September 2025, the agreement committed the company to destroying the original files it had downloaded from piracy sites, but explicitly did not require destruction of the trained models themselves. Round Hill is asking for something more aggressive: impoundment of "copies retained in training datasets, internal repositories, model weights, and server infrastructure".

This distinction matters because model weights are the mathematical parameters that make a trained AI model function. Destroying weights would essentially render the model unusable. Round Hill is also demanding a complete accounting of all training data, scraping activity, and datasets involving its musical compositions. The complaints cite Judge William Alsup's June 2025 ruling in the Bartz v. Anthropic case, which stated: "There is no carveout, however, from the Copyright Act for AI companies".

How Is Round Hill Building Its Case Against Anthropic's Claude?

Round Hill's complaint against Anthropic relies heavily on evidence from Anthropic's own published materials. The company released a human-feedback training dataset called hh-rlhf on Hugging Face in 2022, and Round Hill alleges this dataset shows Claude returning copyrighted song lyrics during testing. In one example cited in the complaint, Claude responded to a Disney songs prompt by quoting from "Let It Go." In another, when asked to build a short story from the lyrics to "Eleanor Rigby," Claude complied, though the response was logged as rejected.

Round Hill treats each Claude model release as a separate act of copying. The complaint names Fable 5 and Mythos 5 (both released June 9, 2026), Sonnet 5 (June 30, 2026), and Opus 5 (July 24, 2026) as distinct instances of infringement. The filing also invokes Anthropic's own policy statements about model distillation. When Moonshot AI allegedly distilled Anthropic's Fable model, Michael Kratsios, director of the White House Office of Science and Technology Policy, wrote that "large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable." Sarah Heck, Anthropic's head of public policy, replied: "Illicit, adversarial distillation is IP theft and industrial espionage that supports adversary military and intelligence capabilities." Round Hill uses this exchange to argue that Anthropic itself recognizes the harm of model extraction.

Sarah Heck, Anthropic's head of public policy

"Anthropic deliberately trained Claude to commit copyright infringement," the complaint states, citing the examples of copyrighted lyrics appearing in Claude's responses.

Round Hill Music, copyright complaint filed August 2026

What Role Did Data-Scraping Tools Play in the Alleged Infringement?

Round Hill's complaint against Anthropic includes detailed allegations about the specific technical choices the company made when ingesting data. According to the filing, Anthropic tested at least three text-extraction programs before selecting one: Readability, Newspaper, and jusText. Round Hill claims jusText was rejected because it left copyright notices and ownership details intact, which Anthropic treated as "boilerplate" and "useless junk." Newspaper was chosen instead for its ability to strip that information.

The complaint against Bright Data, the data-scraping firm that provided services to Suno, makes a similar argument about deliberate removal of copyright management information. Round Hill alleges that Bright Data's Scraping Browser renders pages without headers and footers, where copyright notices typically appear. The filing argues: "Making the deliberate choice to exclude headers and footers is identical to making the deliberate choice to remove CMI," referring to copyright management information.

Steps Round Hill Is Taking to Strengthen Its Legal Position

• DMCA Violations: Both complaints include Digital Millennium Copyright Act (DMCA) counts stacking on top of per-work statutory damages. Under DMCA Section 1203, Round Hill seeks up to $2,500 for each act of circumvention and up to $25,000 for each removal of copyright management information, multiplying potential liability.
• Scienter Standard Arguments: Round Hill cites Stevens v. CoreLogic to establish that removal of copyright information must be intentional and that defendants must have known, or had reasonable grounds to know, it would induce, enable, facilitate, or conceal infringement. An October 2025 Concord ruling held similar allegations sufficient at the pleading stage.
• Licensing Market Evidence: The complaints name ElevenLabs, Musical AI, Symphonic, Soundverse, GEMA through PLAI, GCX/Rightsify, and Troveo as evidence of a functioning licensing market that the defendants bypassed. PLAI launched on July 23, 2026, under a month before these filings, and GEMA won a first-instance case against Suno on July 31, 2026.

What Are the Specific Numbers in These Lawsuits?

The $1 billion figure in both complaints is a projection contingent on Round Hill amending its exhibits to cover thousands more works. Currently, Exhibit A in each case lists 500 musical compositions, though Round Hill asserts rights in 16,873 recordings and calls the 500-work exhibit a bellwether it will expand.

The only fixed arithmetic in either filing appears in the contributory infringement count against Bright Data: 500 works at $150,000 each, totaling $75 million. Round Hill also cites Bright Data's own financial performance, putting the company past $300 million in 2025 revenue, growing 50 percent year-over-year. The filing alleges the services Bright Data provided to Suno "are only good for copyright infringement".

For context, Round Hill's allegations against Suno are informed by hacked source code reported by 404 Media on July 15, 2026, which included a dataset logged as youtube_music holding over two million music clips and 113,879 hours of audio, roughly thirteen years of continuous music. Round Hill also notes that Suno's guardrails appear inconsistent, claiming a prompt naming Reba McEntire is blocked while a misspelling of her name is not.

What Happens Next in These Cases?

Round Hill has stated its intention to take both cases to trial. The Suno filing runs to 24 pages, while the Anthropic complaint extends to 26 pages. Neither allegation has been tested in court, and the defendants have not yet filed responses. Suno, Anthropic, and Bright Data were contacted for comment at the time of reporting but did not respond.

The legal strategy employed by Round Hill differs markedly from settlement-focused approaches in previous AI copyright disputes. By seeking impoundment of model weights themselves, Round Hill is testing whether courts will require not just the destruction of training data, but the functional elimination of trained models. This represents a potential inflection point in how copyright law applies to generative AI systems.