Music rights firm sues Anthropic and Suno over copyrighted training data

The Reservoir Lawsuits

Reservoir Media Management has filed two separate lawsuits against the AI companies Suno and Anthropic. The legal actions, filed on Oct. 21, allege that these companies used copyrighted music and lyrics to train their artificial intelligence systems. The firm is seeking a combined $338 million in damages from the two entities.

The lawsuit against Suno specifically targets the platform's use of 1,050 copyrighted songs. These include well-known tracks such as Ring of Fire by Johnny Cash, Both Sides Now by Joni Mitchell, and Papa Don’t Preach by Madonna. Reservoir claims Suno scraped this content to power its tool, which allows users to generate new songs from text prompts.

In the second lawsuit, Reservoir claims Anthropic used lyrics from 853 songs to train the Claude family of AI models. The list of songs in this filing includes Big Yellow Taxi by Joni Mitchell, Lush Life by Billy Strayhorn, and Bring Me to Life by Evanescence. Anthropic responded to reports by stating they disagree with the allegations and intend to defend the case vigorously.

Allegations of Data Scraping

The lawsuits claim that the AI companies bypassed restrictions like logins and paywalls to access protected content. Reservoir also alleges that the companies stripped away essential metadata, including titles, writer credits, and publisher information. These actions are being framed as clear instances of copyright infringement by the music rights company.

Regarding Anthropic, the filing suggests the company used a text-extraction tool called Newspaper. This tool reportedly removes copyright notices from written material during the data collection process. Additionally, the lawsuit claims Claude was partly trained on lyrics from Common Crawl, a web crawler that may have picked up data from licensed sites like Musixmatch and LyricFind.

For Suno, the allegations involve scraping music from various platforms, including YouTube Music, Deezer, and the lyrics database Genius HQ. A previous hack of Suno's code revealed a folder containing more than 2 million clips from YouTube Music. The lawsuit argues that Suno was designed to replace original artists by providing AI-generated substitutes to listeners.

The Mechanics of AI Training

AI models are developed by feeding vast amounts of data into a system to help the software learn patterns. This data can include text, images, music, videos, and software code. By processing these inputs, models learn to generate new content that can mimic human-created styles, such as writing a song or creating an image.

The training process often relies on large-scale web crawling to gather information from across the internet. When these crawlers pick up copyrighted material, it can lead to legal disputes over whether the use constitutes fair use or theft. The current legal landscape is struggling to define the boundaries of how much data a company can ingest without permission.

Suno and Anthropic are part of a larger trend where multiple companies are facing scrutiny for their data sources. Other major models like GPT, Gemini, and Grok have also been accused of ingesting mountains of data. This has led to the creation of trackers to monitor the growing number of copyright infringement lawsuits against the AI industry.

A Growing Legal Landscape

Anthropic is not the only company facing these types of accusations. The firm was sued last week for allegedly using 274 Elvis Presley songs to train Claude. This follows a history of litigation for the company, including a $1.5 billion settlement regarding claims that it trained its AI on millions of stolen books.

Suno is also facing pressure from other major industry players. The company is currently involved in copyright infringement lawsuits from both Sony Music and Universal Music Group. These high-profile disputes highlight the tension between the desire for technological progress and the need to protect the creative work of human songwriters.

Reservoir CEO Golnar Khosrowshahi stated that the music industry cannot allow AI platforms to exploit the work of songwriters. She noted that the company will hold platforms legally accountable to ensure rights are upheld. The goal is to establish a framework that supports new technology while protecting the creative work at the heart of the music industry.