Elvis Rights Holders Sue Anthropic for $41M Over AI Training
The owners of Elvis Presley's catalog sued Anthropic for training Claude on 274 hit songs, demanding up to $41 million in damages.
Policy · Source: Music World (Billboard report)
What happened
The owners of Elvis Presley's music are taking Anthropic to court. A unit of Authentic Brands Group filed a lawsuit on October 23 claiming the AI company trained Claude on 274 of the rock legend's copyrighted songs. The plaintiffs allege Anthropic used these tracks without permission to build a business model based on massive copyright infringement at a scale never seen before. Several other entities that own Elvis rights joined the filing as co-plaintiffs.
The lawsuit targets the unauthorized use of iconic hits. The list includes Can't Help Falling in Love, Don't Be Cruel, All Shook Up, Return to Sender, and A Little Less Conversation. The plaintiffs are demanding up to $150,000 per infringed song under the Digital Millennium Copyright Act. That brings the total potential damages to around $41 million. They also want every unlicensed copy destroyed, including those kept in training datasets and model weights. Anthropic has not yet commented on the lawsuit.
This case shares DNA with other recent legal battles. Lead attorney Richard Busch just filed a nearly identical lawsuit against Anthropic on behalf of Reservoir Media over Johnny Cash and Madonna songs. Busch is known for winning the massive Blurred Lines copyright case for Marvin Gaye's estate. Now he is applying that same aggressive strategy to artificial intelligence. The complaint argues Anthropic's plan was simply to infringe now and hope for a two-trillion-dollar IPO later.
Key facts
- 274 — Number of copyrighted Elvis songs Anthropic allegedly used
- $150,000 — Maximum statutory damages requested per song
- $41 million — Total potential damages at stake in the lawsuit
- October 23, 2026 — Date the lawsuit was filed
Why it matters
This is the highest-profile AI lawsuit filed on behalf of a single artist. AI companies rely heavily on the fair use defense to scrape the internet for training data. They argue that training a model is like a human reading a book to learn. Rights holders disagree. If courts decide that ingesting copyrighted lyrics and audio is infringement rather than fair use, the cost of building foundation models will skyrocket. Startups will have to license data before training, creating a massive barrier to entry for anyone without billions in the bank.
The second-order effect is a massive shift in power from tech companies back to legacy IP holders. We are already seeing AI companies sign licensing deals to avoid court. Suno recently settled with Warner Music Group and signed agreements with BMG. If Anthropic loses or settles for millions, every major estate, record label, and publisher will line up to extract their pound of flesh. The era of free training data is closing. Foundation models will become heavily regulated, licensed products rather than open frontiers.
For builders
Audit your training data pipelines immediately
You can no longer scrape the web blindly and hope for the best. If you build AI products, you must track exactly where your training data comes from. Founders who ignore data provenance risk fatal lawsuits from aggressive IP holders. Keep clean records of what enters your models.
Budget for expensive data licensing deals
The fair use defense is under severe stress in federal courts. Smart builders are already paying for proprietary data access instead of scraping. You need to factor data licensing costs into your financial models today. Relying on free scraped data is becoming a massive legal liability that investors will flag.
Implement strict output guardrails in your apps
Your models cannot spit out exact copies of copyrighted work. You must build filters that block users from generating derivative works of famous IP. If your tool generates a knockoff Elvis song or recites copyrighted lyrics, you will be the one paying the fine. Guardrails are now a core product feature.
My take
Tech companies got arrogant. They thought they could scrape the entire history of human creativity for free and hide behind the fair use doctrine. Richard Busch does not lose music copyright cases. Anthropic is about to learn that building a trillion-dollar business on stolen intellectual property comes with a heavy price tag. Pay the creators, or get sued into oblivion.
Original reporting: Music World (Billboard report). This is my rewrite and opinion.