Elvis Presley Rightsholders Sue Anthropic Over 274 Songs

The owners of Elvis Presley's music have sued Anthropic, the maker of Claude, in a complaint filed on Friday, October 23, 2026. According to Rolling Stone, which credits Billboard with the first report, the case is captioned ABE EPE LLC v. Anthropic and alleges that the company infringed 274 Presley songs. The plaintiffs seek $150,000 per track, which Rolling Stone puts at roughly $41 million. Every allegation below is the plaintiffs' claim; no court has ruled, and Anthropic had not commented when the reports ran.

This is the latest in a growing line of music-copyright cases against Anthropic, and the first one in our coverage brought by the rightsholders of a single artist's catalog. If you build with or on top of large language models, it is worth understanding what is actually claimed, what is merely alleged, and what it could mean for the way models handle song lyrics.

Quick facts: the Elvis v. Anthropic lawsuit

What the complaint alleges

Two separate theories run through the quotes that both outlets published. The first is about training: the plaintiffs say Anthropic used Presley's recordings and compositions to build Claude without a license. The second is about output: they say Claude will reproduce or closely rework the songs on request.

Lead attorney Richard Busch wrote that the plaintiffs "bring this action to enforce these exclusive rights against the defendant that has built its entire business model based on copyright infringement at a scale never seen before." He added that they "are deeply protective of these iconic songs."

The output allegation is the more specific one. The complaint says Claude "has created infringing output of copyrighted material of virtually every song from every catalog available on the internet, including the Works." It then says that even after Anthropic "supposedly put up 'guardrails' after being sued," a recent request to recreate specific songs "for a modern audience" still produced what the plaintiffs describe as near-copies.

The complaint quotes Claude as saying the renders "keep landing very close to the originals: same structure, same hooks reused with light rewording, and full sets of verses/choruses/bridges." We have not seen the underlying chat log, only the complaint's quotation of it, so treat that exchange as the plaintiffs' exhibit rather than an independently verified transcript.

Rolling Stone adds that, like the "Blurred Lines" litigation, the suit argues Anthropic is using Presley's songs to create "unlicensed derivative works" with its models. That framing matters. A pure training claim leans on the fair-use question; a derivative-works claim points at what the model emits.

Why $150,000 per song, and where does $41 million come from?

The number is not a surprise to anyone following these cases. US copyright law lets owners of registered works elect statutory damages instead of proving actual losses, and the ceiling per work is $150,000 when a court finds infringement willful. We unpacked that math in our explainer on why Sony and Warner want $150,000 per song from Anthropic.

Multiply 274 by $150,000 and you get $41.1 million, matching Rolling Stone's "roughly $41 million." Two caveats keep that figure in perspective:

• It is the maximum theoretical exposure on the pleaded works, not a prediction. Courts often award far less than the ceiling, and a finding of willfulness is required for the top tier.
• It only counts the 274 listed songs. The plaintiffs' broader claim is that Claude infringed "virtually every song," but damages in a statutory case attach to specific registered works.

A copyright stamp above a page, standing for the Anthropic music copyright lawsuit over unlicensed training

How this fits the wider run of Anthropic lawsuits

The Hollywood Reporter places the Elvis case in a sequence. Round Hill sued Suno and Anthropic in August, alleging that hundreds of its songs were used to train their models without permission. Earlier, Universal Music Publishing, Abcko, Concord and BMG sued Anthropic over lyrics being used to train Claude. For the Sony and Warner filing, see our original report on the late-August complaint.

Rolling Stone characterizes the Elvis suit as one of the first major instances where an artist's estate, rather than a label or trade group, has taken legal action against an AI company. The company has been pursued on several other content fronts too: Canadian news publishers sued over articles, and a $1.5 billion book-piracy settlement was approved in July.

The industry-wide picture is larger still. Our write-up of an AI lawsuit tracker covering 248 cases against 76 companies shows copyright claims dominating, and USA Today's $250 million suit against OpenAI shows the same playbook applied to text.

Music notes floating out of a box, illustrating song lyrics and melodies generated by an AI model

Training versus output: the legal question that matters

Reading the lawsuit's two theories together explains why music cases may be harder for AI companies than some book cases.

Training. Anthropic has previously argued that training on copyrighted material can be protected by US fair use. Courts have not given one uniform answer, and the book-piracy settlement showed that how data was acquired can matter as much as how it was used. The Elvis complaint does not appear, in the reporting we found, to hinge on pirated acquisition; it says the training itself was unlicensed.

Output. A song is short, instantly recognizable, and a few lines of lyrics can be a recognizable fraction of the work. If a plaintiff can show a model reproduces whole verses or tracks structure closely, the output claim does not depend on winning the training argument at all. That is why this complaint spends effort on a specific Claude exchange.

For a broader sense of how models handle lyrics, our post on the Granny benchmark and Claude's lyrics behavior tests how well refusals hold up against indirect requests, which is exactly the area the Elvis plaintiffs point to when they say guardrails fell short.

What is confirmed, and what is not

Readers who want to check for themselves can pull the docket once it is available on a public court-records service, or read Billboard's original report. We could not access the Billboard article directly in our research, so the details above rest on Rolling Stone and The Hollywood Reporter, both of which attribute the first report to Billboard.

What this means for people who build with Claude and other models

Nothing in the filing changes Claude's availability or pricing. But the pattern is relevant to builders in three ways:

1. Product design. If your app lets users ask a model to "rewrite this song for a modern audience," you may be building the exact feature the plaintiffs call infringing. Consider filtering requests that name a specific copyrighted song and ask for a close rewrite.
2. Provenance records. If you fine-tune on music, lyrics or any scraped text, keep a record of where each source came from and under what license. The legal exposure in these cases often turns on acquisition and willfulness, which provenance records help you answer.
3. Vendor risk. Check whether your AI vendor offers indemnification for copyright claims on outputs, and what it excludes. See our AI regulation guide for the policy context across jurisdictions.

For teams that run coding or content agents with broad permissions, AgentBeam, the agent security platform from the explainx.ai team, is one way to review what an agent is about to do before it ships content you did not intend. It is a security tool, not a copyright filter, so it complements rather than replaces a licensing review. You can learn more at AgentBeam.

What to watch next

• Anthropic's answer. A motion to dismiss or an answer will show whether the company leans on fair use, on the quality of the infringement allegations, or on both.
• Consolidation or coordination. With multiple music-publisher suits pending against the same defendant, courts may be asked to coordinate overlapping issues.
• Guardrail changes. Companies often tighten lyric handling after high-profile complaints. Watch for changes to how Claude responds to requests to recreate or "modernize" named songs.
• Settlement or licensing. The industry has already seen licensing deals between AI music companies and rightsholders. Whether Anthropic pursues a licensing route here is an open question we cannot answer from the reporting.

Bottom line

The Elvis Presley lawsuit puts a famous catalog and a specific, quotable output allegation in front of a court. The money figure, about $41 million, is a ceiling the plaintiffs chose, not a prediction. The real stakes are the two legal questions underneath: whether training on songs without a license is permitted, and whether a model that rewrites a song "for a modern audience" is producing a derivative work. Until Anthropic responds and a judge rules, treat both outlets' summaries as the plaintiffs' account.

Details are accurate as of October 25, 2026 and may change as the docket develops. This post is news coverage, not legal advice.

Related reading

• Why Sony and Warner want $150,000 per song from Anthropic
• Sony and Warner sue Anthropic over song lyrics in training data
• Anthropic's $1.5 billion book-piracy settlement approved
• Canadian news publishers sue Anthropic
• AI lawsuit tracker: 248 cases against 76 companies
• USA Today sues OpenAI for $250 million
• Granny benchmark: Claude, copyright and lyrics