Sony Music Entertainment filed a new lawsuit against Udio in a New York court on Monday, alleging that the AI music generator infringes the copyrights of more than 30,000 of Sony’s songs. The complaint, as reported earlier by Music Business Worldwide, lists tracks spanning iconic artists including Elvis Presley’s “Hound Dog,” Beyoncé’s “Say My Name,” and Harry Styles’ “As It Was.”

Sony’s filing also frames that list as the tip of a bigger problem, arguing that it represents “only a small portion of Plaintiffs' works that Udio infringed.” In other words, the number is meant to sound finite, but the claim is not: Sony is telling the court that the infringement footprint is larger than the already huge catalog it named.

This is the third-party AI music industry at full legal volume. Udio is part of a wave of tools that generate songs from prompts, but the music licensing question sits at the center of why labels keep turning to court. These lawsuits are not just about individual tracks. They are about the training data pipeline, which determines what models “learn” from, what can be reproduced, and what evidence labels will be able to present when they argue infringement.

Sony’s move follows a similar pattern from 2024, when Sony, along with Universal Music Group and Warner Records, filed a lawsuit against Udio and the AI music-maker Suno. That earlier case matters because it describes how labels believe they can substantiate their claims. In the source, Sony is said to have gained access to Udio’s training data through discovery. That detail is important for executives because it points to a strategy that is procedural, not only creative: labels can use legal process to request information that private companies do not readily volunteer.

The new lawsuit widens the scope again. It explicitly asserts infringement across a massive set of works, and it ranges from older standards like Elvis Presley material to modern mainstream hits like the Harry Styles track named in the complaint. That spread is more than a brag list. It signals that Sony is targeting the idea of “style” and “recognizability” with a concrete anchor: specific songs the label says are protected works.

From a board-level perspective, these cases are also about leverage. AI music companies and the platforms around them face a dual risk: (1) direct legal exposure, and (2) uncertainty that chills partnerships. Even if an AI music tool is popular, a label lawsuit can complicate collaborations with rights-holders, distribution partners, and advertisers who do not want to be caught in the blast radius of alleged infringement. The practical consequence is that more legal friction can slow down scaling, even before any court decision.

There is also a regulatory-adjacent reality to consider, even when the fight is currently in court rather than in a regulator’s office. Copyright law is the governing framework for works and training datasets in the United States, but enforcement actions often shape how companies build compliance. The second-order effect is that AI music providers may increasingly invest in documentation, dataset controls, and provenance systems, not because regulators demanded it yesterday, but because discovery risk and litigation costs make “trust us” an expensive strategy.

The stakes extend beyond Sony and Udio. If labels can keep stacking allegations across tens of thousands of songs, it sets a precedent for how future complaints might be framed: large representative lists, paired with arguments that those lists are only a portion of the alleged infringement. For executives at other AI music companies, content platforms, and even adjacent players like music discovery apps or creator tooling, the message is clear: the legal conversation will not stay narrow.

For decision-makers in the broader creator economy, the strategic takeaway is uncomfortable but actionable. These lawsuits treat AI music generation as something that must be defended at the level of specific copyrighted works, not just general concepts of originality. The more Sony expands the named catalog in a New York filing, the more it pressures the entire AI music market to confront the same question: where does the model’s learning come from, what evidence supports that it is not copying, and who pays if the answer is litigated to the end.