Sony and Universal file new lawsuit against Suno over 60,202 recordings

Background of the renewed legal action

On 18 September 2026, Sony Music and Universal Music filed a new complaint against the AI music‑generation platform Suno. The filing marks a continuation of a dispute that began with earlier allegations concerning the use of copyrighted audio in training data. The complaint explicitly references a set of alleged infringements that the majors claim represent only a fraction of the material they consider to be at issue.

According to the complaint, the two record companies allege that Suno copied at least 60 recordings that are protected under copyright law. The figure of 60 recordings is presented by the plaintiffs as a subset of a larger corpus that they assert has been used without permission. No additional numbers are provided, and the complaint does not disclose the total size of the dataset in question.

The majors also contend that Suno circumvented YouTube’s protection mechanisms by employing download tools. In a prior procedural context, Suno had acknowledged that it had obtained audio data from YouTube using the open‑source utility yt‑dlp. This admission forms part of the basis for the claim that the platform deliberately bypassed technical safeguards to harvest copyrighted material.

Technical arguments surrounding model v6

A central point of the complaint concerns Suno’s most recent model, designated v6, which was launched on 9 September 2026. The plaintiffs argue that v6 continued to benefit from learning derived from earlier models, specifically through the aggregation of user preference data when two tracks were presented as alternatives. The complaint suggests that this mechanism allowed the newer model to inherit knowledge that originated from the disputed dataset.

In contrast, Suno maintains that v6 was trained from scratch using a distinct set of data sources. The company asserts that the training set for v6 does not contain any recordings owned by Sony or Universal, thereby rejecting the allegation that the new model derives value from the contested material. This position creates a factual divergence that the pending litigation must resolve.

The development of v6 involved collaborations with other music rights holders, namely Warner Music, BMG, and Believe. These partners participated in the creation of the model and are reported to receive a share of the revenue generated from its use. The involvement of these entities is documented in the launch announcement, and it demonstrates that Suno has pursued licensed relationships with some major players while remaining in dispute with others.

Potential remedies and financial stakes

Sony and Universal’s complaint seeks both an injunction to halt the distribution of v6 and monetary damages. The damages sought are quantified at up to $150,000 per alleged infringing work, as reported by the media outlet Next. This figure represents the maximum amount per work that the plaintiffs claim could be awarded, though the actual award would depend on a court’s assessment of liability and damages.

The request for an injunction reflects the majors’ desire to prevent further exploitation of the contested model while the case is pending. An injunction, if granted, could stop Suno from offering v6 to users, thereby impacting the platform’s revenue streams and its partnerships with Warner Music, BMG, and Believe.

The financial claim of $150,000 per work is presented in the complaint as a potential ceiling for damages. The actual calculation of damages would require a detailed analysis of the alleged copying, the market impact, and any statutory damages applicable under copyright law. No definitive amount has been adjudicated, and the figure remains a claim rather than a court‑determined sum.

Open questions and evidentiary limits

One unresolved issue concerns the distinction between future licensing agreements and alleged liability for past data usage. The complaint separates the question of whether Suno might obtain licenses for future content from the claim that it is already responsible for past infringements. This separation raises legal questions about the retroactive application of licensing obligations.

Another area of uncertainty is the evidentiary basis for the claim that v6 benefitted from prior models via user preference data. The plaintiffs argue that the preference mechanism constitutes a form of knowledge transfer, yet Suno contends that the model was trained anew without such data. The factual determination of whether the preference data effectively re‑introduced copyrighted material into the training process is central to the case.

The admission that Suno used yt‑dlp to download audio from YouTube establishes a factual record of data acquisition. However, the legal significance of that admission depends on whether the downloaded content was protected, whether the download constituted a breach of YouTube’s terms of service, and whether such a breach translates into copyright infringement under applicable law.

• The complaint alleges at least 60 copyrighted recordings were copied
• The claim includes alleged circumvention of YouTube protections using yt‑dlp
• Sony and Universal seek an injunction and damages up to $150,000 per work
• Suno asserts v6 was trained from scratch without Sony or Universal recordings

The pending nature of the litigation means that no judgment has yet resolved these competing factual narratives. Until a court renders a decision, the parties’ positions remain assertions supported by the limited evidence presented in the complaint and the company's prior statements.

If the court were to find that Suno’s v6 model incorporated protected recordings, the injunction and damages sought could materially affect the platform’s operations and its relationships with licensed partners. Conversely, a finding that v6 was trained independently could reinforce the viability of AI music generation models that rely on newly sourced data and licensed collaborations.

The broader industry context suggests that the outcome of this case could influence how AI music generators negotiate licensing with major labels, how they manage data provenance, and how courts interpret the role of user‑generated preference data in model training. The case therefore sits at the intersection of copyright law, technology, and emerging business models in the music sector.