Yes, BMG Rights Management filed a landmark copyright infringement lawsuit against Anthropic on March 17, 2026, claiming that the company’s Claude AI system has reproduced verbatim passages from 493 copyrighted musical compositions without authorization or compensation. The lawsuit alleges that Anthropic’s training process included copyrighted songs owned by major artists including Justin Bieber, Bruno Mars, Ariana Grande, the Rolling Stones, and Louis Armstrong, and that Claude can output complete or near-complete lyrics upon request. With potential statutory damages of up to $150,000 per infringed work, Anthropic’s total liability could exceed $70 million.
This article explains the specific allegations, the artists affected, how much money is at stake, and how this lawsuit fits into a broader pattern of copyright disputes involving AI companies. This latest copyright claim comes just months after Anthropic settled a separate lawsuit with authors for $1.5 billion over the use of pirated books in its training data. Unlike the authors’ settlement, which involved unpublished works scraped from illegal sources, the BMG lawsuit focuses on commercially available, protected musical compositions—raising new questions about what constitutes fair use when training AI systems on copyrighted material.
Table of Contents
- What Are the Specific Copyright Allegations Against Anthropic?
- What Is the Potential Financial Liability and Damages?
- Which Artists and Their Work Are Named in the Lawsuit?
- How Does This Music Copyright Lawsuit Compare to the Previous Authors’ Settlement?
- Why Didn’t Anthropic Respond to the Cease-and-Desist Letter?
- What Is the Status of the Bartz Settlement Fairness Hearing?
- What Happens Next, and What Is the Broader Implication for AI Copyright?
What Are the Specific Copyright Allegations Against Anthropic?
BMG Rights Management’s lawsuit identifies 493 separate instances where Claude allegedly reproduces copyrighted musical compositions in violation of copyright law. According to the complaint, when users request song lyrics or musical content, Claude outputs substantial portions of these protected works without permission. The company claims this represents direct copyright infringement under U.S. federal law, which protects original musical compositions regardless of whether they were included intentionally in training data or reproduced as a side effect of how the AI model was built.
The lawsuit names some of the world’s most commercially successful songs as examples of verbatim reproduction. “Uptown Funk” (Bruno Mars), “You Can’t Always Get What You Want” (Rolling Stones), and “What A Wonderful World” (Louis Armstrong) are specifically cited in the complaint as songs Claude can reproduce in full or substantial portions. These aren’t obscure works—they are among the most recognizable songs in popular music, and their inclusion in Claude’s training data appears deliberate rather than accidental. The distinction matters legally: intentional inclusion of protected works is harder to defend as fair use than inadvertent capture of a small sample.
What Is the Potential Financial Liability and Damages?
Under U.S. copyright law, owners can pursue statutory damages of up to $150,000 per infringed work when copyright violation is established. With 493 claimed violations, Anthropic’s potential liability reaches approximately $74 million—though actual damages awarded could be lower if the court finds certain violations less egregious or awards damages on a reduced scale. However, if the court determines the infringement was willful (that Anthropic knowingly and deliberately violated copyright), damages could increase. Conversely, if the court narrows the scope of violations or finds some claims unsubstantiated, damages could be substantially less.
This potential exposure places BMG’s lawsuit in a different category than typical AI copyright disputes. The previous Bartz v. Anthropic settlement with authors reached $1.5 billion—making the authors’ case more costly overall—but that covered a much larger class of individuals. The BMG lawsuit, focused on a single rights holder’s catalog, suggests that other major rights holders (music publishers, film studios, visual artists) may file similar suits, each potentially worth tens of millions of dollars. Anthropic has not publicly disclosed whether it carries insurance that would cover copyright infringement liability, though such policies are common in technology companies.
Which Artists and Their Work Are Named in the Lawsuit?
BMG Rights Management represents rights to compositions from some of the world’s most famous recording artists, and the lawsuit emphasizes this breadth to demonstrate the scale of alleged infringement. The artists explicitly mentioned include Justin Bieber, Bruno Mars, Ariana Grande, the Rolling Stones, and Louis Armstrong—spanning multiple decades and genres of popular music. This diversity matters in copyright litigation: it shows the alleged infringement is not limited to a narrow niche but affects major commercial artists whose songs generate significant licensing revenue.
The three songs specifically highlighted—”Uptown Funk,” “You Can’t Always Get What You Want,” and “What A Wonderful World”—represent different copyright strategies. “Uptown Funk” is a modern hit protected by current copyright holders; “You Can’t Always Get What You Want” is a classic rock composition with long-standing copyright protection; and “What A Wonderful World” is a jazz standard that generates ongoing royalties for rights holders. By citing works from different eras, the lawsuit establishes that Claude reproduces music spanning decades, suggesting the infringement is systematic rather than a one-off technical glitch. This strengthens BMG’s argument that Anthropic had access to—and likely deliberately included—a substantial music corpus in training data.
How Does This Music Copyright Lawsuit Compare to the Previous Authors’ Settlement?
The Bartz v. Anthropic settlement from September 2025 involved a different copyright issue: authors’ objection to Anthropic using pirated books from illegal sources like Library Genesis and Pirate Library Mirror to train Claude. That settlement cost Anthropic $1.5 billion but resolved a class action covering numerous authors whose works were used without permission. The settlement also included a favorable ruling from Judge William Alsup, who determined that using lawfully purchased books for AI training could qualify as fair use, but using pirated works could not. This distinction—between legally acquired versus stolen source material—was central to that case.
The BMG lawsuit presents a different legal theory. Rather than arguing that Anthropic violated copyright by using protected works in training, BMG is arguing that Anthropic violated copyright by enabling Claude to output the protected works themselves. This focuses on the model’s output and capability rather than its training sources. While the authors’ case centered on whether training on certain books was permissible, the music case centers on whether the trained model can reproduce copyrighted content on demand. A court could theoretically find that training on music was permissible fair use (similar to how humans learn songs) while also finding that enabling Claude to reproduce those songs verbatim violates copyright. The legal question is different, and the stakes are different: the authors were seeking damages for unauthorized use in training, while BMG is seeking damages for unauthorized distribution (via Claude’s outputs) of protected compositions.
Why Didn’t Anthropic Respond to the Cease-and-Desist Letter?
BMG sent Anthropic a formal cease-and-desist letter in December 2025, requesting that the company take steps to prevent Claude from reproducing copyrighted musical compositions and offering to negotiate a licensing agreement. Anthropic did not respond to this letter and did not engage with BMG on potential licensing terms, according to the lawsuit. This non-response is significant in copyright litigation: it can be interpreted as either confidence that infringement claims are meritless or indifference to the rights holders’ concerns. Courts sometimes view failure to respond to a cease-and-desist as evidence of willfulness, which can increase damages. The silence may reflect Anthropic’s legal strategy rather than dismissal of BMG’s concerns.
The company may have determined that responding and negotiating could itself constitute an admission of liability, or it may have calculated that the cost of licensing music retroactively was prohibitive. Alternatively, Anthropic may have believed its use of music falls within fair use protections for AI training. However, the lack of engagement meant that no licensing agreement was reached, and BMG proceeded with litigation. This is common in high-stakes IP disputes: when rights holders and technology companies cannot reach settlement early, litigation becomes necessary. A licensing agreement reached in December 2025 might have prevented the lawsuit entirely, but Anthropic’s lack of response foreclosed that option.
What Is the Status of the Bartz Settlement Fairness Hearing?
The Bartz v. Anthropic settlement, which was reached in September 2025, has a scheduled fairness hearing in April 2026. This hearing is a standard requirement in class action settlements, where the court must determine whether the settlement is fair, reasonable, and adequate for the class of authors it represents. At this hearing, the judge will consider whether the $1.5 billion payout fairly compensates authors for unauthorized use of their works in training data, whether the settlement terms are appropriate, and whether any individual authors wish to opt out or object. The timing of this hearing—just days or weeks after the BMG lawsuit was filed—may focus additional media and legal attention on copyright issues in AI.
The fairness hearing could influence how courts view the BMG case. If Judge William Alsup (or another judge) signals during the Bartz hearing that copyright claims against AI companies have merit, BMG’s litigation gains momentum. Conversely, if the court emphasizes fair-use protections for AI training, Anthropic gains use in the music case. The decisions in these two cases will likely shape how other rights holders—film studios, image databases, visual artists—approach their own copyright claims against AI companies. The April 2026 hearing is not just about the authors; it is a pivot point for copyright law in the AI era.
What Happens Next, and What Is the Broader Implication for AI Copyright?
The BMG lawsuit is likely to face a motion to dismiss, where Anthropic will argue that copyright claims should be rejected on legal grounds (such as fair use or lack of direct infringement). If the motion is denied, the case will proceed to discovery, where BMG will request evidence of how Claude was trained, what data sources were used, and what mechanisms allow Claude to reproduce copyrighted songs. This discovery phase could last months and will generate significant public scrutiny of Anthropic’s training practices. Settlement negotiations could begin at any point, or the case could proceed to trial in 2027 or later.
The broader implication is that AI companies face a multi-front copyright battle. The authors settled for $1.5 billion; the music industry is now suing for potentially tens of millions more. Visual artists, filmmakers, and other creators are considering similar suits. Unlike copyright disputes from the internet’s early years—when courts ruled that search engines and social media companies had broad safe harbors for user-generated content—copyright suits against AI companies are testing whether those protections apply when companies directly use copyrighted works to train proprietary models. How Anthropic fares in the BMG case will set precedent for whether other tech companies can be sued successfully on similar grounds, and whether licensing agreements become mandatory for AI training.
You Might Also Like
- Class Action Claims Microsoft Copilot Reproduced Copyrighted Journalism From Major Publishers
- Lawsuit Claims Whoop Fitness Tracker Shared Heart Rate Variability Data With Life Insurers
- Lawsuit Claims Thomson Reuters Sold Journalists’ Source Emails to Government Without Warrant