SI Agent News
DevelopingClaude

Anthropic sued over using Elvis Presley music to train Claude AI

Rights holders are seeking millions in damages, alleging the AI company used copyrighted lyrics and songs without permission to develop its models.

US$48.6milTotal damages sought by plaintiffs

The short version

Why it matters

This case highlights the ongoing legal tension between AI development and the protection of intellectual property. It could set a precedent for how 'fair use' is defined when large-scale datasets are used to train generative models.

The Elvis Presley Lawsuit

The owners of rights to hundreds of Elvis Presley songs have filed a lawsuit against the San Francisco-based AI company Anthropic. The legal action was filed on Friday, Oct 23, in a federal court in San Jose to address allegations of copyright infringement. The plaintiffs include several entities, such as Authentic Brands Group and Raleigh Music Publishing, who claim their works were used without permission.

The lawsuit specifically names 274 musical compositions used to train the Claude AI models. These include well-known hits like "All Shook Up," "Return to Sender," and "A Little Less Conversation." The plaintiffs argue that these copyrighted works were integrated into the AI's development process without any authorization from the rights holders.

Anthropic has responded to the allegations by stating they disagree with the claims. A company spokesperson told the press on Monday, Oct 26, that the firm intends to defend the case fully. This legal battle joins a growing list of disputes between the music industry and technology companies over the use of creative content.

Data collection and storage

The plaintiffs allege that Anthropic gathered copyrighted lyrics from across the internet and from massive datasets used for machine learning. The complaint points to specific collections, including The Pile, Common Crawl, and the Internet Archive. They argue that Presley's lyrics were included in these datasets without the necessary permissions from the owners.

Beyond the initial training phase, the lawsuit claims that Anthropic maintained a permanent internal repository of this copyrighted material. The plaintiffs argue this repository could be used for future training and other company purposes. They contend that the copying occurred at multiple stages, including when the material was acquired, cleaned, and stored.

Furthermore, the complaint accuses Anthropic of stripping away important copyright management information. This includes the removal of song titles, credits for songwriters and composers, and official copyright notices. The plaintiffs argue that these actions were taken while incorporating the material into successive versions of the Claude models.

AI outputs and filters

A major focus of the lawsuit is what Claude produces when users interact with it. The plaintiffs claim the AI has generated new lyrics from the Presley catalogue and created derivative works based on the original songs. They argue that the AI is capable of mimicking the specific structures and hooks of the protected music.

The complaint notes that while Anthropic added filters after previous lawsuits, these safeguards can be bypassed. Users can allegedly circumvent these protections by slightly changing the wording of their requests. The plaintiffs cite an instance where Claude created versions of songs for a "modern audience" while keeping the original verse, chorus, and bridge arrangements.

The legal team argues that these outputs retain too many elements of the original compositions. By keeping the core structure and hooks while only changing some of the wording, the plaintiffs believe the AI is infringing on their rights. They are seeking to stop this behavior and have the company destroy unauthorized copies of the works.

Broader industry impact

This case is part of a wider trend of music publishers and copyright owners challenging generative AI. Other major companies, including Universal Music Group and Concord Music Group, have also sued Anthropic over allegations involving copyrighted lyrics. These disputes center on how AI models learn to replicate human creativity from existing data.

Anthropic's legal troubles are not limited to the music industry. The company has also faced litigation regarding its use of copyrighted books. Last year, Anthropic reached a US$1.5bil settlement with a group of authors who accused the company of using pirated books to train its models. These cases highlight the complexity of navigating intellectual property in the age of AI.

The current lawsuit comes at a significant time for Anthropic, as the company recently filed paperwork for a proposed initial public offering. The plaintiffs are asking a judge to prohibit further infringement and order the destruction of unauthorized copies. This includes copies allegedly held in training datasets, model weights, and server infrastructure.

What to watch

Try it

You can check if an AI can bypass its own filters by asking it to write a song in a specific style while slightly varying the prompt wording.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by The Star · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news