Sony Pursues New Claim Against Udio Over 30,000 Recordings
The music giant files a second lawsuit after a court blocked expansion of its original complaint, while rivals Universal and Warner have already signed licensing agreements with the AI startup.
A Strategic Pivot After Court Rejection
Sony has initiated a second legal action against AI music platform Udio, this time centered on 30,117 recordings the label claims were exploited without proper licensing. The decision to file separately follows a June 29 court ruling that rejected Sony's attempt to fold these additional works into an existing 2024 complaint it had originally brought alongside Universal and Warner.
The court's reasoning offers insight into the procedural chess match unfolding: while judges declined to expand the scope of the first case, they explicitly affirmed that rights holders retain the ability to pursue infringement claims and damages across all copyrighted material. Sony appears to have taken that guidance as a green light to launch parallel litigation rather than wait for resolution of the initial matter.
At DailyTechWire, we've tracked similar bifurcation strategies in semiconductor patent disputes across Asia, where plaintiffs often file multiple narrow actions to preserve claims while negotiating. The tactic can create pressure on defendants facing escalating legal costs and uncertainty across product lines.
The Catalog at Stake
Documentation reviewed includes a roster of commercially significant tracks spanning decades: works by Alicia Keys, Dolly Parton, Elvis Presley, Beyoncé, Bob Dylan, Britney Spears, and Michael Jackson. The breadth suggests Sony is asserting claims over marquee assets that generate substantial licensing revenue through traditional channels, from streaming platforms to synchronization deals in film and advertising.
The scale of the alleged infringement carries financial weight. Sony is seeking statutory damages of up to $150,000 per work, a figure that could theoretically approach $4.5 billion if applied to all recordings cited. While courts rarely award maximum statutory amounts, the nominal exposure underscores the leverage Sony wields in any settlement negotiation.
YouTube as Training Ground
The new complaint introduces a technical allegation: that Udio sourced audio for model training by extracting files from YouTube. Udio has previously acknowledged using YouTube audio data, defending the practice as fair use under US copyright doctrine. That defense hinges on whether courts view generative AI training as transformative, a question that remains unsettled across jurisdictions.
Fair use analysis in the US typically weighs four factors: the purpose and character of the use, the nature of the copyrighted work, the amount used, and the effect on the market. Udio's argument likely centers on transformation, the same rationale Google successfully invoked when the Supreme Court ruled in 2021 that copying Java APIs for Android constituted fair use. Yet music holds stronger emotional and commercial value than functional code, and courts have historically granted narrower fair use latitude to creative works.
The YouTube extraction method also raises questions about terms of service. While YouTube's policies prohibit downloading content without explicit permission, enforceability varies. More relevant to Sony's claim is whether Udio's use substitutes for licensed training data, a market Sony and other labels are actively building through deals with AI firms.
The Licensing Divide
Universal and Warner, Sony's original co-plaintiffs, have both reached licensing agreements with Udio. Warner framed its deal as a potential revenue stream for artists who opt in, signaling a pragmatic shift toward monetizing AI rather than blocking it outright. Those agreements likely grant Udio retroactive immunity for past training while establishing payment structures for future use.
Sony's decision to continue litigating sets it apart. In the new filing, the label argues that Udio's willingness to license now only confirms the illegality of its earlier conduct, when it trained models on copyrighted material without seeking permission. The framing is pointed: if licensing was always the proper path, why didn't Udio pursue it from the start?
This divergence mirrors broader strategic splits within the music industry. Some labels view AI music tools as existential threats to session musicians, producers, and the scarcity economics that underpin catalog value. Others see them as distribution channels, akin to how streaming once disrupted physical sales but ultimately expanded total listening. Sony's stance suggests it either believes it can extract better terms through litigation or views deterrence as more valuable than near-term licensing revenue.
Regional Echoes and Export Control Parallels
The Sony-Udio dispute unfolds against a backdrop of intensifying IP enforcement across Asia's AI sector. South Korea's music labels have filed similar claims against domestic startups training on K-pop catalogs, while Japan's Agency for Cultural Affairs has proposed guidelines that would require explicit consent for copyrighted works used in machine learning. China's regulatory approach differs: Beijing issued draft rules in 2023 requiring generative AI services to respect IP rights but has left enforcement largely to civil courts, where outcomes remain unpredictable.
The legal uncertainty around training data echoes debates over semiconductor export controls. Just as US restrictions on AI chips to China have forced firms to redesign products around allowable specifications, unclear copyright boundaries are pushing AI music startups to either pre-license content or risk retroactive liability. The cost of compliance is rising: according to industry estimates we've reviewed, comprehensive licensing for a training dataset covering major label catalogs can run into eight figures, a threshold that advantages well-funded players and risks entrenching incumbents.
What Comes Next
Sony's dual-track litigation could stretch for years, particularly if Udio contests the fair use question through appeals. The strategic calculus for Udio is straightforward: settle with Sony and lock in certainty, or fight and potentially set precedent that benefits the broader generative AI industry. A win for Udio would embolden other firms to train on copyrighted material without permission; a loss would validate the labels' licensing model and likely increase the price of entry for new competitors.
For Sony, the downside of losing is not just foregone damages but erosion of its ability to control how its catalog is used. If courts rule that training on copyrighted music constitutes fair use, the label loses a key negotiating tool in future deals. That risk may explain why Sony has chosen to litigate even as peers settle.
The outcome will ripple beyond music. Publishing, film, and software industries are watching these cases closely, as they grapple with similar questions about whether generative AI training requires licensing. A clear judicial ruling, in either direction, would provide the certainty that stakeholders across creative industries have been seeking since generative models began proliferating in 2022.
