Field noteNov 14, 20265 min read1 source

Labels oppose Suno's full stay in UMG II, propose a partial stay with targeted discovery on Suno v6

In one minute

Record-label plaintiffs in a new copyright case against Suno oppose freezing the entire case until a fair-use ruling in an earlier suit. Their filing proposes postponing ownership discovery on more than 60,000 recordings while allowing narrower inquiry into newer models, licensing, business expansion and finances. The court has not ruled; the filing frames a dispute over how to preserve evidence as models change.

What changed: The labels oppose a full stay in the new Suno case and seek targeted discovery, including scrutiny of Suno v6, while deferring work-specific ownership discovery until the earlier case rules on fair use.

What happened

In a memorandum filed in the District of Massachusetts, plaintiffs including UMG Recordings and Sony Music Entertainment respond in partial opposition to Suno’s motion to stay the case. They agree that discovery confirming ownership and chain of title for the more than 60,000 recordings asserted in this action should wait until the court rules on fair use in the earlier case, *UMG Recordings, Inc. v. Suno, Inc.* They object to pausing all activity, which they say could leave the new action frozen through a decision expected realistically into 2028.

The proposed alternative has three parts. First, ownership discovery on the newly asserted recordings would be deferred. Second, the plaintiffs seek limited discovery on four subjects: Suno’s successor models and ongoing model development, including training data, methods and provenance for 2025 and 2026 models such as v6; Suno’s efforts to license training data; commercial expansion, including new markets, distribution channels, features and funding; and Suno’s financial condition after the earlier case’s discovery record closes. The plaintiffs say they would provide reciprocal information about licensing recordings for AI training and the effect of AI-generated music on the market for their works.

Third, the plaintiffs propose that documents, written discovery responses and deposition testimony from the earlier case be treated as produced or taken in this action too, subject to a protective order. They argue that this would limit duplication while allowing discovery of recent conduct not covered in the earlier record. They describe their plan as a workable starting point and say the parties could ask the court to adjust its scope if disputes arise or more discovery proves necessary.

The filing situates the request in the two cases’ different records. The plaintiffs say the earlier case’s document collection largely stops at February 2025, interrogatory responses cover models through v5, and corporate testimony reaches through v5.5 with only limited v6 topics. They state that Suno released v6 on September 9, 2026, and that the new action was filed on September 18. Allegations in the complaint—not findings by the court—say v6 was trained in part on user interactions, uploads and outputs from earlier models, and that techniques including knowledge distillation carried learned behavior forward. The filing also says the earlier case has a November 20, 2026 fact-discovery deadline and dispositive motions due June 22, 2027.

The plaintiffs argue that waiting years could make evidence harder to reconstruct as engineers leave, memories fade and training pipelines change. They cite their recent learning that an engineer responsible for data scraping and training-data acquisition had left Suno. Their filing emphasizes that the proposed limited discovery would address recent model and business activity while deferring the much larger task of establishing ownership of the asserted recordings. It is advocacy in a pending motion, not a judicial finding that the allegations are true or that discovery should proceed on those terms.

What the archive adds

New Runtime’s earlier coverage includes a November 9, 2026 note on Suno’s answer in a separate copyright case, which reported Suno’s admission that it obtained audio from YouTube for training while denying many allegations and asserting fair use. The present filing is a new procedural development in a related dispute: plaintiffs seek targeted discovery on newer models while conceding that ownership discovery can wait.

Trend signal

This development connects to agent security moves to runtime boundaries only weakly: the dispute concerns model-training provenance and operational records, rather than runtime controls for agents. The more direct signal is that AI companies’ evolving training practices and model lineage can become difficult-to-examine evidence when product versions change faster than litigation proceeds. This does not yet fit a pattern New Runtime tracks.

Why it matters

For teams building generative products, the case illustrates why versioned records of training data, model changes, licensing activity and ownership of source material matter beyond engineering reproducibility. A product may move through several generations while discovery or compliance review remains anchored to an earlier snapshot. If records are scattered or the people who built a model leave, reconstructing what happened can become harder precisely when questions about provenance become consequential.

The filing does not establish that Suno infringed copyright, that v6 contains outputs from earlier models, or that the plaintiffs’ proposed discovery is proportionate. Still, its requested categories offer a concrete view of the evidence a rights dispute may turn on: model lineage, data acquisition, licensing efforts, commercialization and financial context. Teams can treat those categories as a prompt to keep contemporaneous documentation and clear version boundaries, without assuming that this particular case determines what other companies must disclose.

Limits and uncertainty

The document is the plaintiffs’ memorandum opposing Suno’s request for a stay, so its account of Suno’s conduct, the risks of delay and the proposed discovery is one side’s position. Allegations about v6’s training sources and model-transfer techniques are not established findings here. The court has not ruled on whether to grant a full stay, adopt the proposed partial stay or set different discovery limits. The filing also does not resolve the fair-use question in the earlier case or the merits of the claims concerning the recordings asserted in this action.

Reader takeaway

The immediate development is procedural, not a copyright ruling: the labels want the largest ownership inquiry deferred, but seek discovery now into Suno’s newer models and recent business. For builders, the practical lesson is to preserve model and data lineage as the product evolves; a record that can be reconstructed later is part of operational readiness.

Open question

Will the court allow targeted discovery into Suno’s newer models while the fair-use issue in the earlier case remains unresolved?

Evidence / sources

  1. 1storage.courtlistener.com/recap/gov.uscourts.mad.306399/gov.uscourts.mad.306399.37.0.pdf
All Field Notes

Next reads

Read next

Open the graph
  1. 01archiveField NotesOpen the latest editorial analysis.
  2. 02sourcesSourcesSee the original sources this archive cites.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregate...

Open the JSON contract