Staff Reports
The owners of Elvis Presley’s song catalog have just filed a federal lawsuit claiming Anthropic trained its Claude AI on 274 Elvis compositions and lyrics without permission. This week’s filing in the Northern District of California is more than another tech-versus-creator dust-up — it tests whether big AI companies can treat copyrighted art like free data. Spoiler: they shouldn’t get a pass.
What the lawsuit says about Anthropic and Claude
The complaint, brought by ABG EPE IP LLC and several rights holders tied to Elvis Presley’s catalog, accuses Anthropic of copying lyrics and musical compositions into internal training libraries and model artifacts. Plaintiffs say Claude can spit out outputs that are near-derivative of classics like “Can’t Help Falling in Love” and “Don’t Be Cruel.” The suit seeks injunctive relief, an accounting of datasets, destruction of unauthorized copies, and statutory damages — plaintiffs plead the $150,000-per-work willful-infringement ceiling that totals roughly $41.1 million as a damages ceiling for 274 works.
Why this case matters for creators and AI
This isn’t just an Elvis headline. It’s one of the first high-profile suits that centers on a single artist’s catalog and targets how a major AI company trains its models. The plaintiffs also claim Anthropic stripped copyright management information and kept copyrighted material in ways that let Claude reproduce or approximate the original songs. If true, that isn’t innovation — it’s appropriation. Courts will have to decide whether training on copyrighted music like this is protected as fair use or whether creators get the protections the Copyright Act intended.
What Anthropic faces next — and why discovery will matter
Expect the fight to turn on internal documents: crawl logs, dataset inventories, model weights, and cleaning scripts. Plaintiffs already demand preservation and a full accounting. Anthropic has said it “disagrees with the claims and intends to defend this case fully,” which is the standard Silicon Valley reply when a lawsuit threatens to pull back the curtain. That means we’ll see motions to narrow discovery, fights over trade-secret claims, and maybe early bids for injunctive relief. Whatever the outcome, this case will shape how AI companies treat music, lyrics, and other creative works going forward.
No free pass for machine learning by theft
There’s a simple principle at stake that should unite left and right: creators earn a living from their work, and the law exists to protect that. If AI companies want to build models on the culture we all love, they should pay for it or find lawful ways to license it. Treating songs like public scrap paper because some engineer wrote a clever scraper is not progress — it’s freeloading. Courts should force transparency and accountability, not let companies treat lawsuits as a cost of doing business. The King deserves better than to be used as training data without a check — and so do every songwriter and artist who isn’t famous enough to make headlines when they’re ripped off.
