Authentic Brands sues Anthropic over 274 Elvis Presley songs

Copyright litigation against generative AI developers has followed a recurring pattern since 2023, typically initiated by rights holder collectives or major publishers rather than individual creators. These suits generally allege that model training constitutes unauthorized reproduction and creation of derivative works, challenging the applicability of fair use defenses in the context of large-scale data ingestion. Prior cases have frequently involved class actions representing thousands of authors or musicians, seeking to aggregate claims to achieve viable economic thresholds for litigation. A consistent procedural feature is the demand for maximum statutory damages per work, leveraging the cumulative potential liability to incentivize settlement or licensing negotiations before substantive fair use rulings occur. Historically, defendants have responded with motions to dismiss based on transformative use arguments, though few cases have reached final judgment on the merits. This pattern reflects a strategic effort by rights holders to establish licensing norms through litigation leverage rather than awaiting legislative clarification, creating a cycle of filing, discovery disputes, and confidential settlements that shapes industry practices incrementally.

On October 23, 2026, Authentic Brands Group (ABG) filed a federal copyright infringement lawsuit against artificial intelligence developer Anthropic, alleging the unauthorized use of 274 songs recorded by Elvis Presley to train generative AI models. The complaint, which was first reported by Billboard on October 25, 2026, identifies ABG as the licensing organization holding ownership rights to both Presley’s music catalog and likeness. According to the filing, Anthropic allegedly ingested these copyrighted works without permission or licensing agreements to develop its Claude AI models. The lawsuit specifically cites iconic tracks such as "Can’t Help Falling in Love," "A Little Less Conversation," and "Don't" as examples of the infringed material.

Authentic Brands Group is seeking statutory damages of $150,000 per song, a figure that aggregates to a total potential liability of approximately $41 million. This calculation relies on the maximum statutory damages available under U.S. copyright law for willful infringement, suggesting the plaintiffs intend to argue that Anthropic’s alleged conduct was not merely negligent but deliberate. Multiple sources confirm that ABG is leading a coalition of plaintiffs in this action, indicating that while the current public reporting focuses heavily on the Presley catalog, the legal action may involve broader representation of rights holders. The Guardian reports that ABG accuses Anthropic of infringing on the rights of the 274 specified songs, framing the suit as a defense of the "King's" intellectual property against unlicensed technological exploitation.

As of the latest available reporting, Anthropic has not issued a detailed public statement addressing the specific allegations regarding the Elvis Presley catalog. However, one source indicates that Anthropic disputes the claims and intends to defend itself, though this assertion appears in a social media post rather than an official press release or court filing. The absence of a formal rebuttal from Anthropic leaves the technical specifics of their training data sourcing unverified in the public record. It remains unclear whether Anthropic utilized the specific sound recordings, the underlying musical compositions, or lyrical transcriptions, a distinction that carries significant weight in copyright adjudication. Several Bluesky posts summarizing the event note that the accusation involves using copyrighted lyrics to train models and produce infringing versions, yet the primary news coverage emphasizes the training aspect.

The timeline of events shows a rapid escalation from filing to public awareness. The lawsuit was formally submitted on Friday, October 23, 2026. Within two days, major entertainment and technology trade publications had picked up the story, with Billboard breaking the news of the specific song count and financial demands. By October 25, outlets including The Guardian, Rolling Stone, and The Hollywood Reporter had published detailed accounts of the litigation. This swift media cycle underscores the heightened sensitivity of the AI industry to copyright disputes involving high-profile cultural assets. The involvement of Authentic Brands Group, a major brand management company, signals that rights holders are increasingly organizing through corporate entities rather than individual estates to pursue AI developers.

The allegations center on the premise that the ingestion of copyrighted music for model training constitutes infringement absent a license. Authentic Brands Group contends that Anthropic’s use of the 274 songs was unauthorized, implying that no fair use defense or existing mechanical license covers the specific application of training large language models. The plaintiffs’ demand for $150,000 per work suggests they are pursuing a strategy of maximizing statutory exposure to pressure a settlement or establish a deterrent precedent. Conversely, the defense’s position, as briefly referenced in secondary sources, rests on disputing the factual basis of the claims, though the legal theory for this dispute has not yet been articulated in public filings accessible via the provided sources.

This case represents a distinct vector in the ongoing tension between generative AI and intellectual property law. Unlike previous disputes involving visual art or code, this action targets the intersection of musical composition, sound recording, and lyrical content within the context of text-based or multimodal model training. The specificity of the claim—274 songs, $41 million, named tracks—provides a concrete test case for how courts may quantify harm in AI training scenarios. As the litigation proceeds, the industry awaits Anthropic’s formal response and the court’s initial rulings on motions that could define the scope of discovery into training datasets.

Authentic Brands Group risks failing to establish licensing leverage if fair use prevails, while Anthropic faces $41 million in direct exposure plus potential injunctive relief affecting model availability. The broader AI industry watches for rulings that could validate or undermine the $150,000-per-work statutory damage framework applied to training data. Music rights holders collectively hold billions in catalog value dependent on AI licensing outcomes; a plaintiff victory here could trigger cascading filings across catalogs. Conversely, defendant success would strengthen fair use precedents protecting computational analysis. Users face potential service disruption if models require retraining or content filtering. Legal costs for both sides will likely exceed seven figures regardless of outcome, diverting resources from development and licensing innovation.