Anthropic Sued Over 274 Elvis Presley Songs in AI Training Lawsuit
The owners of 274 Elvis Presley songs have sued Anthropic for tens of millions of dollars, alleging that Claude used copyrighted music for training and produced infringing outputs.
The owners of 274 songs made famous by Elvis Presley have filed a federal lawsuit against Anthropic, accusing the artificial intelligence company of using the musician's catalog without permission to train its Claude models. Seeking tens of millions of dollars in damages, the plaintiffs argue that Anthropic exploited protected creative works to develop commercial generative technology. Anthropic has disputed the allegations and confirmed that it intends to mount a legal defense.
Allegations over Claude's musical outputs
The legal complaint asserts that Anthropic's alleged infringement went beyond merely ingesting copyrighted lyrics into training datasets. According to the lawsuit, Claude was able to generate responses that closely resembled existing Elvis songs, even after Anthropic implemented internal safeguards intended to prevent the reproduction of copyrighted material.
The plaintiffs claim that prompts asking the AI to recreate or adapt songs for modern listeners produced versions that retained recognizable elements of the original works. While AI developers frequently argue that analyzing patterns across large datasets constitutes legally protected fair use, rights holders argue that systems generating unauthorized adaptations and near-identical outputs cross the line into direct infringement. The case highlights the unresolved legal tension between learning from existing content and generating commercial outputs that closely mimic protected expression.
Expanding legal pressure across the AI industry
The dispute adds Anthropic to a growing list of generative AI vendors facing resistance from the music business. Major music publishers have already targeted Anthropic in prior litigation regarding the alleged unauthorized ingestion and output of copyrighted song lyrics. Beyond text-based models, specialized AI music generators like Suno have also faced legal challenges from record labels over their training datasets.
Because Presley died nearly 50 years ago, the case also demonstrates how deeply AI training practices intersect with historic, high-value intellectual property catalogs. AI firms have routinely argued that licensing every piece of training data would be financially and logistically prohibitive, comparing data ingestion to technological progress. However, rights holders argue that commercial industries like television and film routinely pay licensing fees to use protected music, and technology companies should not be granted special exemptions from copyright rules simply because their tools operate via complex software.
What it means for developers
For engineering teams and enterprises deploying foundation models into production, this lawsuit points to potential liability risks surrounding generative outputs. If an underlying model can be prompted into generating text, lyrics, or creative structures that closely track protected intellectual property—even when safeguards are active—applications built on top of those APIs could inadvertently serve infringing content to end users.
Developers building creative assistants, content generation tools, or rewrite workflows should avoid relying solely on the model provider's default filtering. Implementing secondary, application-side guardrails and strict prompt controls can help prevent models from outputting recognizable creative fragments associated with third-party intellectual property.
The expanding legal battles also highlight the risks of tying an entire product stack to a single model provider. If copyright litigation forces providers to alter training sources, retrain models, or restrict specific generation capabilities, single-vendor applications face sudden operational vulnerabilities. To hedge against vendor disruption, developers can evaluate and switch between providers using services like Apixo (https://apixoai.online), which provides affordable, pay-per-token API access to Claude, GPT, Gemini, and DeepSeek through a unified interface. Adopting a model-agnostic approach allows technical teams to maintain uptime and adapt their architecture as the legal landscape around AI training data continues to evolve.
Source: Anthropic sued for allegedly using 274 Elvis Presley songs in AI training — and it's about time AI companies… — TechRadar AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
