Sony opens a larger front in the AI music copyright battle
Sony Music Entertainment has filed a new lawsuit against AI music company Udio, accusing the platform of infringing the copyrights of more than 30,000 songs. The filing dramatically expands the scale of Sony’s claims and raises the stakes in one of the most important legal fights over how generative AI systems are trained and commercialized.
According to the report, Sony says the works identified in the new case range from classic recordings such as Elvis Presley’s Hound Dog to modern hits including songs associated with Beyonce and Harry Styles. The company argues that the list represents only part of the material it believes Udio copied and ingested into its generative AI models.
The new complaint follows a prior round of litigation launched in 2024, when Sony joined Universal Music Group and Warner Records in suing Udio and fellow AI music company Suno. That earlier case had a much smaller formal scope. A judge rejected Sony’s motion to add the much larger song list to the original case, leaving that suit centered on 333 works. Sony has now turned the broader set of claims into a separate action.
Why the new filing matters
The scale alone makes the lawsuit significant. Moving from hundreds of works to more than 30,000 changes the legal and economic pressure on the defendant. Sony is reportedly seeking an injunction to stop infringement and statutory damages of up to $150,000 per work. Even allowing for the usual gap between a claimed maximum and any final award, the numbers underscore how copyright litigation can become existential for AI companies whose products depend on massive training datasets.
Just as important is the procedural signal. Sony says it identified the additional recordings after obtaining access to Udio’s training data in discovery and applying an audio fingerprinting technique. That suggests the case is not only a policy dispute about fair use or innovation incentives. It is also becoming a technical dispute over traceability, dataset provenance, and the evidentiary tools rightsholders can use to connect model training to specific copyrighted recordings.
For the wider AI industry, that is a crucial development. Many generative systems were built during a period when collection at scale often moved faster than legal clarity. If courts accept increasingly granular methods for linking model inputs to copyrighted source works, companies may face greater pressure to document licenses, filtration methods, and data-handling practices in far more detail than they once expected.
The case against Udio
Sony’s complaint, as described in the source material, argues that Udio copied and ingested protected sound recordings into its AI models. The company also points to a prior statement attributed to Udio acknowledging that its generative systems were built by showing the program a vast quantity of sound recordings, including material from YouTube.
That allegation goes to the center of the dispute. AI music generators do not merely store songs in a consumer-facing library; they learn statistical patterns from large collections of recordings and then generate outputs in response to prompts. Copyright law has not fully settled how that process should be treated when the underlying works were not licensed. Rights holders argue that copying for training is itself infringement and that the resulting systems can threaten the market for the original works. AI developers often emphasize transformation, technological progress, and the distinction between training inputs and generated outputs.
The new Sony lawsuit does not resolve that debate, but it does sharpen it. By attaching a vast catalog of specific songs to the complaint, Sony is making the conflict more concrete. It is no longer only an abstract argument about whether AI music models learned from copyrighted recordings. It is a fight over which recordings, how many, and under what conditions they entered the system.
An industry split is emerging
One of the more revealing details in the report is that Universal Music Group and Warner Music Group have already settled with Udio and are now partnering with the company. That marks a notable divergence inside the music business. Rather than maintaining a single litigation front, major labels appear to be taking different positions on whether AI music firms should be fought, licensed, partnered with, or some combination of all three.
This split reflects the broader industry tension around generative AI. Entertainment companies see both threat and opportunity. On one side is the risk that AI systems are built on unlicensed material and can flood markets with imitative works. On the other is the possibility of new revenue streams, new production tools, and new licensing structures if AI companies can be brought into a controlled commercial framework.
Sony’s new suit shows that the rights-enforcement path is still very much alive, even as some peers move toward business arrangements. That divergence could shape the next phase of negotiations between content owners and AI firms. If some companies settle while others escalate, the market may end up with a patchwork model in which access to catalogs depends on a mix of litigation pressure, licensing leverage, and company-specific strategy.
A defining test for generative media
The lawsuit is also a test case for a broader question reaching far beyond music: how much of the generative AI economy rests on training practices that courts will ultimately reject or constrain. Music is an especially potent arena because recordings are commercially valuable, technically identifiable, and tied to a long-established licensing system. That makes it easier to frame a complaint and harder to dismiss the dispute as purely theoretical.
If Sony succeeds in expanding legal exposure around training data, the implications could extend into film, publishing, visual media, and other areas where copyrighted works have been used to build AI products. If Udio prevails or reaches terms that normalize large-scale licensing, that could instead accelerate the shift toward a negotiated marketplace for training rights.
For now, the practical takeaway is clear: the legal fight over generative AI music is not narrowing. It is widening, becoming more specific, and increasingly tied to evidence extracted from the systems themselves. Sony’s decision to file a separate lawsuit over more than 30,000 songs signals that the copyright battle around AI-generated media is entering a more consequential phase.
This article is based on reporting by The Verge. Read the original article.
Originally published on theverge.com
