Elvis Presley Rights Holders Sue Anthropic Over Alleged AI Training on His Music

Elvis Presley’s rights holders have reportedly sued Anthropic, alleging that the artificial intelligence company used Presley’s copyrighted music to train its AI systems without permission. The reported lawsuit involves Anthropic’s Claude model or related systems and raises a central question in the dispute over music copyright and AI training: can technology companies use protected songs to develop commercial AI models without obtaining licenses?

The Guardian reported that the alleged use involved 274 Elvis Presley songs. Other summaries describe similar claims that Anthropic trained its models on Presley’s music without authorization. Source 5

The allegations have not been established by a final court ruling. The available summaries do not identify the complete complaint, court, case number, requested damages or Anthropic’s formal response. Those details will shape the dispute.

The case could affect more than Presley’s catalog. It may influence how courts evaluate AI training on copyrighted music, whether music publishers can demand licensing fees from AI companies and how developers document the sources used to build large language models.

What the Lawsuit Allegedly Claims

According to available reports, Presley’s rights holders claim that Anthropic used copyrighted Presley songs to train its AI models without authorization. One summary specifically identifies Claude, while others refer more generally to Anthropic’s models. Source 3

The alleged conduct could involve:

• Collecting copyrighted songs or lyrics.
• Storing or reproducing those works in a dataset.
• Processing the works during model training.
• Retaining copies for development, testing or evaluation.
• Generating lyrics or other protected material in response to users.

These activities are not legally identical. A training claim may focus on whether copying occurred when the works entered the training pipeline. A separate claim could concern whether the model later reproduced lyrics or other substantially similar expression.

A system can discuss Presley, identify a song title or summarize a recording without reproducing protected lyrics. Conversely, a system that generates lengthy or accurate lyrics could raise different copyright concerns.

The available material does not establish precisely how the songs were obtained, whether they appeared in a specific dataset or how Anthropic processed them.

The Reported Use Involves 274 Songs

The Guardian reported that the alleged AI training involved 274 Presley songs. Source 5 Other summaries repeat that figure. Source 9

The available sources do not provide the complete list of songs, the alleged copying dates, the datasets involved or whether all 274 works were used in the same way. Individual titles should not be attributed to the lawsuit without reviewing the original complaint or another official filing.

Who Controls the Relevant Rights?

The phrase “Elvis Presley’s rights holders” may describe companies, estates, publishers, record labels or other organizations with contractual authority. A Presley song may involve separate rights in:

• The musical composition.
• The lyrics.
• The sound recording.
• Licensing arrangements held by publishers or labels.
• Publicity or personality rights.
• Commercial uses of Presley’s name and image.

An AI-training claim could concern the composition, lyrics, sound recording or several rights at once. Ownership may differ from one song to another. A company that controls sound recordings may not control the underlying compositions, while an estate or licensing company may have commercial authority without owning every copyright interest.

The exact plaintiffs named in the complaint are necessary for a precise ownership analysis. Plaintiffs generally must establish ownership, exclusive rights or contractual authority before pursuing an infringement claim.

Why AI Training Creates a Copyright Dispute

Rights holders may argue that AI training requires the collection, storage, processing or reproduction of copyrighted works. If protected songs entered a training dataset without authorization, they could claim that Anthropic’s development process implicated their exclusive rights.

Music is particularly complex because lyrics, melodies, arrangements and recordings may involve separate protections. Rights holders may argue that commercial AI development using large music catalogs should require licenses rather than bypass established licensing markets.

That argument does not automatically establish infringement. A court would need to examine the actual copying, purpose, works involved, model-development process and available defenses.

Fair Use Could Become Central

The available summaries do not confirm Anthropic’s legal strategy, but fair use could be important. Courts generally consider four factors:

1. Purpose and character. Anthropic could argue that training serves a technological or analytical purpose. Rights holders could respond that the use is commercial and supports a profitable product.
2. Nature of the work. Songs are highly creative works, which generally receive strong copyright protection.
3. Amount used. The court may examine whether Anthropic used entire songs, substantial excerpts or limited portions.
4. Market effect. Rights holders may argue that AI training harms current or potential markets for music licenses. Anthropic could argue that training does not substitute for consumers’ ordinary use of songs.

The court may also consider retained copies, lyric reproduction, output behavior and whether a practical market for AI-training licenses exists. No universal rule currently resolves every dispute involving copyrighted AI-training data.

Potential Effects on Claude and Other AI Models

The lawsuit could bring closer scrutiny to Anthropic’s data practices. Evidence may include dataset records, acquisition agreements, internal policies, filtering systems, technical documentation and model-testing results.

The analysis may differ depending on whether material came from public content, licensed databases, users, third-party data collectors or unauthorized copies. A ruling against Anthropic could increase pressure on AI developers to negotiate licenses and maintain detailed records of training-data sources.

The case would not automatically determine the legality of every other AI system. Training methods, datasets and retention policies vary across companies.

Output-Related Concerns

Training claims and output claims are separate, although they may overlap. The court could examine whether Claude can reproduce lyrics, continue copyrighted text, provide extensive excerpts or generate passages closely resembling protected lyrics.

The available reports focus on alleged training use, not specific Claude outputs involving Presley songs. There is no verified basis in the supplied material to state that Anthropic reproduced particular Presley lyrics.

If a model reliably produces protected lyrics, rights holders could argue that the behavior provides evidence about training, memorization or infringement. Anthropic could distinguish model development from individual user requests, depending on the facts and claims.

Possible Effects on the Music Industry

The dispute could encourage rights holders to seek paid AI-training licenses covering catalog size, duration, purpose, geography, output restrictions, attribution, reporting, audits and compensation.

Licensing could create new revenue for rights holders while giving AI companies clearer permission to use music. The terms could vary substantially between a large catalog used for a commercial foundation model and a small research dataset. Rights holders may also seek safeguards that prevent models from producing lyrics or other substantial protected expression.

The case could increase pressure on AI companies to:

• Maintain detailed dataset inventories.
• Record how copyrighted works were acquired.
• Remove disputed works from future training.
• Create rights-holder opt-out systems.
• Negotiate licenses before collecting large catalogs.
• Block requests for full song lyrics.
• Preserve technical records for litigation.

The available reports do not establish whether Anthropic has adopted or announced any specific response measures.

Key Legal Questions

Did Anthropic Copy Protected Works During Training?

The case may turn on evidence showing how Presley songs were obtained, stored and processed. Relevant evidence could include training datasets, acquisition records, internal policies, development documents and technical tests.

Claude’s ability to discuss Elvis Presley would not, by itself, establish unauthorized copying. The plaintiffs would likely need to connect the songs to Anthropic’s training process and show that the relevant use violated rights they control.

Was the Use Transformative or Commercially Substitutive?

Anthropic could characterize training as a technological use that analyzes patterns rather than distributes songs. Rights holders could argue that commercial AI development exploits expressive works without compensation and threatens a new licensing market.

The court may examine whether outputs substitute for licensed content and whether AI-training licenses represent a market that rights holders could reasonably control.

Could the Case Influence Other AI Disputes?

A music case could affect disputes involving books, news articles, images, scripts, television programs and recorded performances. The broader impact would depend on the court’s reasoning, because different works involve different datasets, copying methods and outputs.

The dispute should not be described as nationwide precedent unless a court issues a significant ruling with broader legal authority. A settlement could resolve the parties’ claims without answering the wider questions.

Timeline and Current Status

Available reporting states that Presley rights holders sued Anthropic over alleged unauthorized use of Presley music in AI training. The reported scope is 274 songs. Source 1

One supplied summary attributes a report to The New York Times dated October 25, 2026. That date requires verification because the underlying article and court filing were not provided. Source 7

The supplied summaries do not identify the court, case number, filing date, causes of action, requested remedies or Anthropic’s formal response.

The likely next steps include Anthropic responding to the complaint, preliminary motions, discovery into training data and model development, and evaluation of ownership, copying, defenses and damages. The parties could settle before trial. A settlement might include payment, licensing terms, data restrictions, output safeguards or confidentiality provisions.

Court filings and official company statements will provide the strongest evidence for future updates.

Conclusion

Presley’s rights holders reportedly claim that Anthropic used 274 Presley songs to train AI models without permission. The dispute concerns whether copying creative works for AI development is permitted, requires a license or violates copyright law.

The outcome remains unresolved. Key issues include whether protected works were copied during training, whether fair use applies, whether rights holders can demand compensation and whether AI companies must provide clearer information about training data.

Until the complete complaint, Anthropic’s response and technical evidence are available, the alleged use and the 274-song figure should remain attributed claims rather than established findings.

Frequently Asked Questions

What do Presley’s rights holders allege?

They allege that Anthropic used copyrighted Elvis Presley songs to train its AI systems without authorization. Available reports state that the alleged use involved 274 songs.

Did Anthropic train Claude on 274 Presley songs?

The reported lawsuit allegedly concerns Claude or related Anthropic systems and 274 Presley songs. The claim requires confirmation through the complaint and other official filings.

Is it illegal to train AI on copyrighted music?

Not automatically in every circumstance. The answer may depend on copying, purpose, the creative nature of the works, the amount used, market effects and defenses such as fair use.

What rights could be involved?

The dispute could involve copyrights in compositions, lyrics and sound recordings, as well as related commercial interests. The exact rights depend on the plaintiffs and works identified in the filing.

What could happen if the rights holders win?

Possible outcomes include damages, an injunction, restrictions on using disputed works, licensing requirements or changes to data practices. The available summaries do not state the requested relief.

Will the lawsuit affect other AI companies?

It could influence how courts, rights holders and developers approach copyrighted training data. Its broader impact will depend on the evidence, the court’s reasoning and whether the case ends in a settlement or published ruling.