A legal battle in a federal court is quietly redefining the boundary between proprietary trade secrets and public judicial transparency in the artificial intelligence era. At the center of this dispute is whether generative music developers must disclose the exact quantity of copyrighted files ingested to build their systems. The resolution of this specific question will establish a powerful precedent for how training data is monitored and regulated.
On May 29, 2026, Suno, Inc. petitioned the United States District Court for the District of Massachusetts to seal a critical metric from public view. This filing responded directly to an opposition letter submitted on May 22, 2026, by Matthew Russell Lee on behalf of Inner City Press. Universal Music Group and Sony Music Entertainment are actively suing Suno for copyright infringement, making the transparency of the training process a matter of immense public interest.
TL;DR Suno and Udio have petitioned federal courts to seal the total volume of training files used in their AI models. Arguing that dataset sizes are proprietary trade secrets, the developers seek to block public access, while transparency advocates and major music labels demand full disclosure during this ongoing copyright litigation.
What is the Model Training Figure and why is it contested?
The legal controversy in Boston centers on a single numerical data point. Suno calls it the Model Training Figure. This value represents the total quantity of audio files used to train its generative models. Confidentiality remains paramount here. Suno insists that keeping this figure secret is vital for its commercial survival.
To support this claim, Suno submitted a formal declaration from its Chief Technology Officer, Georg Kucsko. The developer argues that releasing this metric would allow competitors to reverse-engineer its proprietary training density and data optimization methods. Rivals could exploit this information.
“In the rapidly evolving and highly competitive generative AI market, the size of a company’s training corpus reflects technical development decisions and strategic judgments concerning model design and performance”
If a competitor obtains this figure, they could benchmark their own systems with high precision. This would allow them to replicate Suno’s developmental path at a fraction of the cost. This represents an unfair advantage. Suno argues that because copyright claims focus on specific identified tracks rather than total training volume, the public does not need this figure to evaluate the case.
Matthew Russell Lee of Inner City Press challenged this protective stance by arguing that the public interest remains paramount. Lee asserted that the central legal question is “among the most consequential legal issues of this era.” He urged Chief Judge F. Dennis Saylor IV to deny the sealing request. Public transparency is essential here. Judicial outcomes will impact millions of creators.
To Sell His AI, Suno’s CEO Must First Convince You That Making Music Is “Not Enjoyable”
Forensic data audits expose the scale of ingestion
The transition from broad policy debates to precise litigation metrics occurred during the discovery phase. Plaintiffs gained access to Suno’s private back-end servers and executed audits of the training corpus. They deployed a proprietary audio fingerprinting technology called Audible Magic to scan the database. This software identifies files by matching their unique mathematical acoustic signatures. The results were immediate.
The results of this audit provided concrete proof. Originally, the labels sued over 560 copyrighted works. Following the audit, the proposed amended complaint seeks to add 61,026 specific recordings. The jump is staggering. This represents a massive escalation in the scope of alleged infringement. This deep inspection has turned a generalized dispute into a highly localized mathematical verification.
The music labels argue that Suno has systematically ingested their catalog without authorization. Suno does not dispute using copyrighted material but relies on a fair use defense. However, the exact scale of the ingestion remains a primary point of contention. The litigation takes place against a backdrop of wider industry shifts, where platforms like Suno represent a disruption to traditional music streaming models and licensing norms. The legal battle continues.
Udio coordinates a parallel defense in New York
Suno is not alone in its pursuit of corpus secrecy. On June 1, 2026, Udio filed a parallel motion in the United States District Court for the Southern District of New York. The developer asked Judge Alvin K. Hellerstein to redact what its lawyers call the Training Data Number. This figure is the total volume of files that the plaintiffs claim Udio used to build its model. The motives are identical.
Udio’s co-founder and CEO, Andrew Sanchez, submitted a supporting declaration. Sanchez stated that disclosing this figure would expose strategic decisions about model structure and development. Like Suno, Udio argues that its competitors would use this information to build rival systems faster. Secrecy protects their market share.
However, Udio faces distinct legal challenges under the Digital Millennium Copyright Act. Sony Music and other plaintiffs claim that Udio bypassed YouTube’s technological locks to download content in bulk. The plaintiffs allege that Udio used a downloading program called YT-DLP to bypass YouTube’s rolling cipher. On April 15, 2026, Judge Hellerstein denied Udio’s motion to dismiss this claim, and Udio subsequently admitted to using YT-DLP. This was a critical turning point.
Commercial valuations collide with massive legal liability
The pressure to maintain corporate secrecy is tied directly to the financial stakes. Suno has achieved a valuation of 5.4 billion dollars, doubling its market worth in just seven months. The startup recently secured 400 million dollars in a funding round led by Bond Capital. With over two million active subscribers, the developer is on track to earn 300 million dollars in annual revenue. The business is thriving.
These high valuations clash with the potential statutory penalties of the lawsuit. Under the Copyright Act, willful infringement can carry damages of up to 150,000 dollars per work. If the plaintiffs prove willful infringement for all 61,026 tracks, the potential liability reaches 9.15 billion dollars. This mathematical reality represents a significant threat to Suno’s capital reserves. The risk is astronomical.
To mitigate this risk, some developers have pursued licensing. Warner Music Group reached a settlement and entered into a commercial agreement with Suno in November 2025. Similarly, Universal Music Group settled its claims against Udio in October 2025. However, licensing negotiations between Suno and the active plaintiffs have stalled. Litigation remains the final path.
How will these secrecy disputes shape future AI regulation?
Federal courts face a critical task in balancing competitive protection against public access. Chief Judge Saylor IV has scheduled the deadline for summary judgment motions in the Suno case for January 8, 2027. This timeline positions the Boston lawsuit behind other major copyright disputes. Federal judges in California and New York are scheduled to hold hearings on similar fair use questions throughout 2026. Other precedents will arrive first.
These upcoming rulings will establish the legal standards for training metadata. If courts permit developers to hide dataset sizes, future plaintiffs may find it harder to audit models. Conversely, requiring full disclosure could expose proprietary methods to foreign and domestic rivals. The implications are profound.
The immediate decision on unsealing the Model Training Figure will set the rules of engagement. Industry observers must watch whether courts treat dataset size as a trade secret or as public evidence. The outcome will signal how transparent the next generation of artificial intelligence development must be. This debate is far from over.
Sources & Further reading
Timeline & Procedural Milestones
- October 2025: Universal Music Group reaches a settlement with Udio. (Source)
- November 2025: Warner Music Group settles its disputes with both Suno and Udio. (Source 1, Source 2)
- April 15, 2026: Court denies Udio’s motion to dismiss the DMCA claim. (Source)
- May 21, 2026: Plaintiffs file Motion to Amend and original impoundment request. (Source)
- May 22, 2026: Matthew Russell Lee (Inner City Press) files an opposition letter regarding unsealing. (Source)
- May 29, 2026: Suno files formal statement to seal the size of its AI training data. (Source)
- June 1, 2026: Udio files a motion to seal its training data metrics, citing competitive harm. (Source)
- January 8, 2027: Deadline for summary judgment motions in the Suno lawsuit. (Source)
Litigation Metrics & Financials
- 560 tracks: Original number of copyrighted works asserted in the Suno lawsuit. (Source)
- 61,026 tracks: Sound recordings proposed in the plaintiffs’ amended complaint against Suno. (Source)
- 30,442 tracks: Sound recordings proposed in the amended complaint against Udio. (Source)
- Up to $150,000 / work: Maximum statutory damages under the US Copyright Act. (Source)
- $9.15 billion: Suno’s maximum potential statutory liability (61,026 works × $150,000 maximum damages).
- Suno Corporate Growth & Funding: (Source for below metrics)
- $5.4 billion: Current market valuation (doubled over a seven-month period).
- $400 million: Capital raised in the latest funding round led by Bond Capital.
- 2+ million: Active subscriber count.
- $300 million: Projected annual revenue.
Key Entities & Named Figures
- Judicial Presiders:
- Corporate Leadership:
- Tech & Media Elements:
* generate randomized username
- COMMENT_FIRST
- #1 Lord_Nikon [12]
- #2 Void_Reaper [10]
- #3 Cereal_Killer [10]
- #4 Dark_Pulse [9]
- #5 Void_Strike [8]
- #6 Phantom_Phreak [7]
- #7 Data_Drifter [7]
- #8 Zero_Cool [7]
