Elvis Presley copyright holders sue Anthropic over AI training

The legal filing

Companies that hold the rights to the music of Elvis Presley have initiated a copyright infringement lawsuit against the artificial intelligence company Anthropic. The legal action was filed in late October within the U.S. District Court for the Northern District of California. This move signals a growing trend of rights holders seeking to protect legacy catalogs from being used in machine learning processes.

The plaintiffs, which include the Authentic Brands Group, allege that Anthropic used Presley's works without obtaining any prior authorization. They specifically claim that the company stored these works in a central library to train its AI models. This unauthorized storage is the foundation of the copyright infringement claim presented in the complaint.

The lawsuit seeks a total of $41.1 million in damages, which breaks down to as much as $150,000 for each individual song. Additionally, the plaintiffs are requesting a permanent injunction to stop the infringement and the destruction of all unauthorized copies. As reported by The Chosun Ilbo on MSN, the case focuses on the unauthorized replication of a significant portion of Presley's discography.

Scope of the music

The litigation covers a total of 274 songs from the Elvis Presley catalog. This includes some of his most recognizable hits, such as "All Shook Up" and "Can't help falling in Love." By including such high-profile tracks, the plaintiffs aim to demonstrate the scale of the alleged infringement across Presley's most famous works.

The complaint argues that the AI model, known as Claude, was trained on these specific works to generate new content. The plaintiffs claim that when users prompt the AI to recreate these songs for modern audiences, the outputs are not merely inspired by the originals. Instead, they argue the AI produces content that retains the specific structure and choruses of the protected music.

The legal team for the copyright holders argues that these outputs constitute unauthorized derivative works. A derivative work is a new creation that is based on or derived from one or more already existing works. The plaintiffs contend that because the AI replicates these core elements, it exceeds the boundaries of fair use or simple inspiration.

Anthropic's defense

Anthropic has officially denied the allegations made by the copyright holders in the lawsuit. A spokesperson for the company stated, "We disagree with the claims and will vigorously defend against this lawsuit." This response indicates that the company intends to challenge the legal interpretation of how its models process and learn from data.

The company faces a complex landscape of litigation regarding how it handles copyrighted materials. It is currently involved in multiple other lawsuits concerning the unauthorized use of books and music. These cases highlight the tension between the needs of AI developers to access large datasets and the rights of creators to control their intellectual property.

The plaintiffs also took aim at the safety measures Anthropic has implemented following previous legal scrutiny. They argue that the current measures are insufficient to prevent the AI from generating infringing outputs. Anthropic must now prove that its technology can produce useful results without violating the specific rights of the music industry.

Broader industry context

This lawsuit is part of a wider pattern of legal disputes involving major music and publishing entities. Anthropic is currently engaged in copyright disputes with both Universal Music and Concord Music Group over the usage of lyrics. These ongoing battles suggest that the music industry is actively seeking to establish boundaries for AI training.

The company has already faced significant financial consequences from similar allegations in the past. Last year, Anthropic reached a settlement involving a payment of $1.5 billion to resolve a class-action suit by authors regarding the unauthorized replication of books. This massive settlement underscores the high stakes involved in how AI companies source their training data.

The outcome of the Elvis Presley case could influence how other legacy artists and labels approach AI technology. If the court finds that training on these 274 songs constitutes infringement, it may lead to more aggressive licensing requirements. Conversely, a ruling in favor of Anthropic could provide a clearer path for AI companies to utilize existing cultural works.

What comes next

The court will now have to determine whether the way Anthropic trained its Claude model qualifies as a violation of copyright law. The focus will likely be on whether the AI's outputs are substantially similar to the original songs in a way that harms the rights holders. The plaintiffs are looking for a definitive stop to the use of their assets in the AI ecosystem.

The case will also likely involve technical examinations of how the AI processes data to create new content. Experts will need to distinguish between a model that learns general patterns and one that stores and reproduces specific protected sequences. This distinction will be central to the court's final ruling.

The resolution of this case may set a standard for how other iconic figures and musical catalogs are treated in the age of generative AI. It serves as a landmark test for the balance between technological advancement and the preservation of intellectual property rights for creators and their estates.