Reservoir Media sues Anthropic and Suno over copyrighted music training

The new legal filings

Reservoir Media Management, a major independent music publisher, has initiated two separate federal lawsuits against the AI companies Anthropic and Suno. The filings were submitted on October 21, 2026, in the US District Court for the Northern District of California. These cases represent a significant escalation in the ongoing battle between the music industry and generative AI technology.

The lawsuits are distinct in their scope and the specific types of intellectual property they claim were misused. The action against Suno covers 1,050 musical compositions, while the action against Anthropic focuses specifically on the lyrics of 853 compositions. Together, the potential statutory damages for these two cases amount to nearly $340 million.

Reservoir Media Management is the primary plaintiff, along with its UK subsidiary and Shapiro, Bernstein & Co. Limited. The company manages a deep catalog of classic popular music, jazz, and folk. This legal move follows other recent litigation involving both companies from publishers like Round Hill Music and BMG.

Allegations against Suno

The complaint against Suno focuses on the unauthorized use of musical compositions to train its generative AI music models. Reservoir alleges that the platform reproduced, distributed, and created unauthorized derivative works from its catalog without obtaining any licenses. The publisher claims Suno used these works to build its system for creating original songs from text prompts.

Beyond simple copyright infringement, the lawsuit accuses Suno of violating the Digital Millennium Copyright Act. The complaint alleges that the company circumvented digital access controls and removed copyright management information from the works. These practices can lead to additional liability for the AI company beyond the initial theft of the music.

To support these claims, the filing cites evidence from Suno's own leaked source code. Analysis of that code reportedly shows multiple Reservoir-owned compositions appearing directly in the data. These include notable tracks such as "Both Sides Now" by Joni Mitchell and "Ring of Fire" as performed by Johnny Cash.

The case against Anthropic

The lawsuit against Anthropic takes a different approach by focusing on the reproduction of song lyrics rather than the audio recordings themselves. Reservoir accuses the company of copying the lyrics of 853 compositions to train its Claude family of models. The complaint argues that Anthropic used these lyrics as a shortcut to avoid the cost and effort of training its models from scratch.

The filing uses pointed language to describe the company's actions as a violation of copyright law. By using these lyrics, the publisher argues that Anthropic bypassed the necessary permissions required to use protected creative works. This distinction is important because it targets the text-based training data used for Large Language Models (LLMs).

Anthropic has already faced similar litigation from other music publishers regarding the use of copyrighted lyrics in its models. In a previous case involving a broader group of publishers, a federal judge declined to dismiss several of the claims. This suggests that the courts are still weighing the legality of using lyrics for AI development.

Impacted artists and works

The catalog of works affected by these lawsuits includes some of the most recognizable songs in popular music history. The filings specifically name tracks by Madonna, Johnny Cash, Joni Mitchell, and Miles Davis. These include iconic pieces like "Papa Don't Preach" and "So What," showing the breadth of the publisher's holdings.

Joni Mitchell is the most prominently represented artist in the Suno complaint. Of the 1,050 compositions listed in that specific suit, 149 are credited to her. This means roughly one in seven of the total works listed in the Suno case belongs to Mitchell's body of work.

The lawsuit also includes works by Ben Harper, Kings of Leon, and Evanescence. By targeting such a diverse range of genres—from folk and jazz to pop and rock—Reservoir is attempting to establish a broad precedent. They want to ensure that no part of their extensive catalog can be used for AI training without compensation.

The fair use debate

The legal status of using copyrighted music to train AI models remains one of the most actively contested questions in modern intellectual property law. The central dispute is whether this training qualifies as fair use under US copyright law. Fair use is a doctrine that allows for the limited use of protected material without permission under specific conditions.

AI companies generally argue that training models is transformative, meaning it creates something new and different from the original work. Conversely, rights holders argue that the mass reproduction of works for commercial AI development is not the type of use Congress intended to protect. As of October 2026, no US federal court has issued a definitive ruling on this specific issue in the context of music.

The Reservoir lawsuits are part of more than 80 AI-related copyright cases filed since 2023. These cases will eventually help shape the foundational legal questions for the entire industry. The outcome will determine how much control creators have over their work in an era of automated generation.