A group of music-rights holders associated with songs recorded by Elvis Presley has sued Anthropic, alleging that the artificial-intelligence company used 274 protected musical compositions to train its Claude models and that Claude can produce outputs resembling those works. The complaint, filed October 23 in federal court in California, opens another front in the dispute over copyrighted material and generative AI. The allegations have not been tested in court, and no finding of infringement has been made.

A new case with a broad catalog

The complaint caption identifies the case as ABG EPE IP LLC et al. v. Anthropic PBC, case number 5:26-cv-12473, in the U.S. District Court for the Northern District of California. Its first page is marked “Document 1 Filed 10/23/26.” The document is a hosted copy of the filing rather than a court-docket URL; that distinction matters when tracing the source, even though the face of the document supplies the docket number and filing date. The complaint names seven plaintiffs: ABG EPE IP LLC; the Luigi Creatore Trust and its trustees; the Susan T. Aberbach Revocable Trust and trustee; a trust created under the Anne-Rachel Aberbach Family Trust and its trustee; Jason Simeon Aberbach; Abilene Music LLC; and Raleigh Music Publishing LLC.

The plaintiffs say they own or control rights in 274 musical compositions, which the complaint lists in an attached exhibit. The named examples include “All Shook Up,” “Can’t Help Falling in Love,” “If I Can Dream,” “Don’t Be Cruel,” “A Little Less Conversation,” “Return to Sender,” and “(You’re The) Devil in Disguise.” The filing describes the group collectively as “The Elvis Plaintiffs,” but the catalog allegations concern compositions connected to multiple songwriters and rights holders, not a claim that Elvis Presley personally wrote each song. The case is about rights the named plaintiffs say they hold in compositions associated with recordings by Presley and other artists.

The underlying lawsuit was filed on October 23. Its inclusion as a current news item rests on separate, independent reporting the following day, not on stretching a publication window to include the filing date. Rolling Stone reported the case on October 24 and said Billboard first reported the suit. This meets the supplied current-news criterion of first independent reporting on October 24, while keeping the event date distinct from the report date.

What the plaintiffs allege

Training copies and copyright information

The complaint advances several related theories. It alleges direct copyright infringement through the copying and use of the listed works in Anthropic’s training process. It also alleges circumvention of technological protection measures and removal or alteration of copyright-management information, often abbreviated as CMI. In the plaintiffs’ telling, copies were collected, processed and used in model training without a license, and metadata identifying ownership or authorship was stripped. These are assertions by litigants, not established facts about Anthropic’s datasets or the legal status of model training.

The distinction between what is alleged and what is proved is central. A complaint sets out one side’s factual account and legal arguments to initiate a case. Here, the plaintiffs contend that Anthropic’s data pipeline included their works and that its handling of CMI violated federal law. The filing does not itself demonstrate that every composition was present in a training corpus, establish how any particular work was acquired, or resolve whether the alleged processes amount to infringement or circumvention.

Alleged outputs from Claude

The plaintiffs further allege that Claude can generate lyrics or other material reproducing protected works in whole or in substantial part. The complaint describes prompts seeking specific lyrics or songs in the style of artists, and quotes a purported Claude response saying generated “renders” landed close to originals, with similar structure and hooks. The quoted passage appears as part of the plaintiffs’ pleading. It is not an independent technical evaluation, a court finding, or evidence in this reporting record that the model always responds this way.

Rolling Stone summarized the suit as accusing Anthropic of training on 274 songs without permission and repeated the allegation about outputs, while noting that the suit was reported first by Billboard. Its independent report corroborates that the case was filed and describes the plaintiffs’ claims; it does not independently test Claude or verify each song’s presence in training data. That attribution boundary should travel with every description of the alleged reproductions.

What the lawsuit asks for

The requested financial remedy is also more nuanced than a headline amount can suggest. The complaint asks for actual damages and profits or, alternatively, statutory copyright damages of up to $150,000 per work if the infringement is found willful. It estimates the total at approximately $41.1 million for 274 works. That is a request made by plaintiffs, not an award, negotiated settlement or estimate of proven loss. The pleading separately requests damages under provisions concerning technological protection measures and CMI, as well as attorneys’ fees and costs.

Beyond money, the plaintiffs seek declarations that Anthropic violated copyright and related federal provisions, a permanent injunction, and an order to impound or destroy unauthorized copies, including materials they allege remain in datasets, repositories, model weights or server infrastructure. They also request an accounting of training data and scraping activity involving the works. Each remedy would require the court to address legal and factual questions; none follows automatically from filing the complaint.

Rolling Stone likewise describes the requested $150,000 per track figure and approximately $41 million total. The primary complaint is the source for the precise relief language and its qualifications. In particular, “up to” matters: statutory damages have a legal framework and are determined by a court if liability is established. The figure is not a bill already owed by Anthropic.

Why the case matters

The suit brings a distinctive catalog and claimant group into an increasingly visible debate about the relationship between creative works and generative systems. It joins earlier copyright disputes involving AI companies, but it is not the same action as prior music-publisher litigation against Anthropic. The plaintiffs here include an Elvis-associated rights company, publishers, trusts and individuals. The significance is therefore partly procedural and partly commercial: a new set of claimants is asking a court to examine whether and how musical compositions were used in developing a text-generating model.

For rights holders, the complaint signals a push to make control over catalog licensing and attribution relevant to AI training and outputs. For AI developers, it illustrates that disputes can concern more than whether a model produces a recognizable lyric. The pleading challenges alleged copying during data preparation, the handling of ownership information, and downstream outputs. Those different stages raise separate questions that may depend on evidence about the works, the training pipeline, model behavior and applicable law.

But the case does not settle the broader debate. A new complaint is not precedent. It does not establish a general rule that training on copyrighted works is lawful or unlawful, nor does it establish that model outputs necessarily infringe. The legal analysis may turn on facts specific to the works and claimed acts, defenses the defendant may raise, and how courts interpret the relevant statutes. The outcome, if litigated, could matter to both rights owners and AI companies, but any prediction of precedent or industry impact remains uncertain.

What remains unknown

At this stage, the material unknowns include what Anthropic will say in response, what evidence supports the claims about acquisition and use of each work, and how the parties will define or measure any alleged output similarity. The supplied sources establish the filing’s identity, date, parties, asserted 274-work catalog, legal theories and requested remedies. They do not establish liability, damages, the presence of each work in training data, or that any allegedly similar output was caused by a particular training example.

The record available for this draft also has a source limitation: the primary document is a copy hosted outside the court’s own docket system. Its first page displays the court, case number, document number and filing date, but an official PACER-hosted link was not located. Rolling Stone provides the independent date-stamped account, and it attributes its first-reporting information to Billboard. These sources are enough to confirm that a lawsuit was reported and to describe what the complaint says, but they do not substitute for docket monitoring or adjudication.

Nor should the filing be conflated with an earlier lawsuit simply because both involve Anthropic and copyrighted material. The candidate’s supplied recent-coverage list contains no report about this plaintiff group’s October 23 action. The overlap is thematic—the broader copyright and AI-training controversy—not duplication of a covered event. On that basis, and with the October 24 independent report establishing timeliness, the story qualifies as new coverage.

Frequently Asked Questions

Who filed the lawsuit?

Seven named plaintiffs, including ABG EPE IP LLC, music-publishing entities, trusts and individuals, filed the complaint against Anthropic PBC. They describe themselves collectively as the Elvis Plaintiffs and say they own or control rights in the listed compositions.

How many compositions are involved?

The complaint identifies 274 musical compositions in an exhibit. That is the plaintiffs’ asserted list of works at issue; it does not independently prove that all 274 appeared in Anthropic’s training material.

Has Anthropic been ordered to pay $41 million?

No. The complaint requests statutory damages of up to $150,000 per work for alleged willful infringement and gives an approximate total of $41.1 million. No damages award or liability finding is established by the sources reviewed here.

What does the complaint say Claude generated?

The plaintiffs allege Claude can reproduce lyrics or create outputs substantially similar to their compositions and quote a purported model response about outputs landing close to originals. Those statements are part of the complaint and remain contested allegations, not a judicial finding or independent test result.

When is the story considered current?

The complaint says it was filed October 23, 2026. Rolling Stone independently reported it October 24, so the story qualifies based on first independent reporting on October 24 rather than the filing date alone.

References