NEW YORK – Sony Music has filed a second copyright infringement lawsuit against Udio, the AI music generator, asserting that 30,117 sound recordings were copied without authorization to train the company’s artificial intelligence model.
The complaint, filed Sunday in the United States District Court for the Southern District of New York, names Sony Music Entertainment and nine affiliated labels, including Arista Records and LaFace. It arrives as every other major music company has quietly negotiated licensing deals with Udio while Sony has refused to settle.
The new case is a direct consequence of a procedural defeat Sony suffered last month. The company sought to add the 30,117 recordings to its original RIAA-coordinated lawsuit filed in June 2024, but a judge denied the amendment on June 29, citing concerns about scope. That ruling explicitly permitted Sony to file the expanded claims separately, and the label moved within three weeks.
Universal Music Group, Warner Music Group, Merlin, Kobalt, Believe, and the National Music Publishers Association have each reached licensing agreements with Udio since the original suits were filed. Those settlements have resolved the majority of the music industry’s copyright exposure with the platform. Sony’s decision to fight rather than settle now leaves it in an increasingly isolated legal posture, pursuing a startup whose other major creditors have taken payments and moved on.
“Udio’s belated embrace of licensing only underscores the unlawfulness of its decision to copy Plaintiffs’ copyrighted sound recordings, without a license, in the first place,” Sony said in a statement accompanying the filing, as reported by Music Business Worldwide.
The original lawsuit, filed in June 2024 alongside coordinated actions against Udio’s competitor Suno, accused the platforms of ingesting vast libraries of copyrighted recordings to generate music that could mimic professional production quality. Suno has since faced its own escalating troubles. Last month, a hacker gained access to Suno’s source code, which reportedly detailed how the company had scraped millions of songs during its training process, raising fresh questions about how AI music generators were initially built.
The three legal claims in Sony’s new complaint mirror the structure of AI copyright litigation that has become standard across the industry. The first two counts cover sound recordings made after and before 1972, the latter now protected under the Music Modernization Act of 2018. The third count alleges circumvention of digital rights management technologies in violation of the Digital Millennium Copyright Act.
The damages sought are significant. Sony is claiming up to $150,000 per infringed work and up to $2,500 for each circumvention violation, a potential exposure that could reach into the billions if the court were to impose statutory maximums across 30,000 works.
What Sony is not seeking, at least not explicitly, is the kind of outcome most of the industry appears to prefer: a licensing deal that creates a long-term revenue stream from AI music companies rather than a court ruling that might extinguish them. The music business has broadly concluded that AI music generation will exist regardless of the copyright litigation, and that negotiated licensing represents the more commercially rational resolution.
Sony’s isolation in that strategy is significant. Earlier this month, a judge approved Anthropic’s $1.5 billion copyright settlement with authors and publishers over training data used to build its Claude AI models, a landmark resolution that confirmed the copyright exposure AI companies face but left open questions about which fair-use defenses might survive in court.
Publishers have separately pursued Google over Gemini AI training, alleging that Gemini ingested copyrighted books without authorization. That case remains pending. Taken together, the cases form a pattern in which copyright holders use litigation to establish leverage, sometimes converting it into licensing agreements and sometimes pursuing it further.
Without the alignment of Warner and Universal, Sony cannot coordinate a unified industry push toward forced licensing terms. What it can do is continue litigating, accumulate discovery about how Udio’s training data was assembled, and use that information as leverage in eventual settlement talks if any come.
The judge’s earlier refusal to add the recordings to the original case introduces some legal risk. Sony must now pursue two parallel cases against the same defendant in the same court, a posture that could invite arguments about redundancy or judicial economy. Whether that complicates the new case will depend largely on whether the cases are eventually consolidated.
Udio did not immediately respond to a request for comment. The company has previously maintained, in filings related to the original lawsuit, that its AI model does not store copies of copyrighted recordings and that its outputs constitute original creative expression. That defense will face a harder test under the new complaint, which Sony says was made possible by audio fingerprinting technology deployed during discovery in the original case.
