Universal and Sony Sue Suno Again Over the v6 Model

Image credit: Generated for Power JDM The Beat.

Universal Music Group and Sony Music Entertainment have filed a second lawsuit against AI music company Suno, arguing that the company’s newer v6 models still rely on copyrighted recordings allegedly used without permission.

The complaint was filed on September 18, 2026, in the U.S. District Court for the District of Massachusetts. According to reporting by Music Business Worldwide, the labels allege that Suno copied more than 60,000 copyrighted sound recordings to develop the systems behind its music-generation service.

The case is still at the allegation stage. No court has ruled that Suno infringed the labels’ copyrights, and Suno has disputed the claims and said it intends to defend itself.

What the new lawsuit alleges

The labels’ complaint identifies 60,202 sound recordings that they say Suno copied without a license. Universal and Sony claim those recordings were used to develop multiple versions of Suno’s models, including v4, v4.5, v5, v5.5, and the newer v6 model.

The complaint reportedly argues that the recordings are only a portion of the material Suno allegedly copied. The labels say they identified recordings in Suno’s training data through audio fingerprinting during discovery in the earlier case.

The filing also makes a separate allegation involving YouTube. Universal and Sony claim Suno or its agents bypassed YouTube’s anti-downloading technology to obtain recordings for training. That allegation is connected to a claim under the Digital Millennium Copyright Act.

The labels are seeking damages and a court order that would stop Suno from continuing to use their recordings. As reported by Music Business Worldwide, the complaint points to a theoretical maximum of up to $150,000 per work for willful copyright infringement. Applied to 60,202 recordings, that could exceed $9 billion before considering other claims or defenses.

Those figures represent the maximum damages being requested under the labels’ legal theory, not an award that has been issued by a court.

Image credit: Generated for Power JDM The Beat.

Why the second suit matters

This case is separate from the first major-label lawsuit against Suno, which was filed in June 2024. That earlier case focused on a smaller group of recordings. A Massachusetts federal judge later declined to add tens of thousands of additional recordings to the original action, allowing the labels to bring those claims in a separate lawsuit.

The new filing therefore expands the dispute rather than simply repeating the earlier case.

It also targets the newer v6 model. Suno has said that v6 was trained from scratch using licensed music connected to partnerships with Warner Music Group, BMG, and Believe. Suno has also said that the model did not include data from Universal or Sony.

Universal and Sony challenge that explanation. Their complaint argues that v6 was influenced by earlier Suno models that allegedly used unlicensed recordings. The filing characterizes the newer system as “the fruit of the same poisoned tree.”

The labels’ theory is that v6 may have been developed using “user interactions” and preference data from older models. In simple terms, Suno’s systems may record which generated track a user prefers when presented with multiple options. Universal and Sony allege that those preferences were based on outputs from earlier models and could have transferred learned musical characteristics into v6.

The complaint also refers to knowledge distillation. A process in which a newer “student” model learns behavior from an older “teacher” model. The labels argue that this means a new model can still carry forward capabilities learned from disputed training material.

The central question is not only whether copyrighted recordings were copied. It is also whether a company can avoid liability by building a new model on outputs, preferences, or learned behavior from an earlier model that allegedly relied on unauthorized copies.

Suno’s position is expected to focus on its licensing arrangements, its description of how v6 was developed, and its broader argument that training generative AI models can qualify as fair use. Those questions will be addressed through the litigation, and the court has not yet decided them.

Where this fits in the 2026 AI music litigation wave

The new Suno case is part of a wider series of legal disputes involving copyrighted music, artist identity, voice data, and AI-generated content.

In August, Jason Isbell and other musicians filed a proposed class action against Suno that takes a different legal approach. As reported by Reuters and other outlets, that case focuses on right-of-publicity claims and alleged biometric voiceprint violations under Illinois’ Biometric Information Privacy Act.

That lawsuit is not centered on the same copyright theory as the Universal and Sony case. Instead, the musicians allege that Suno used artists’ names, identities, vocal characteristics, and musical personas without permission. The case raises questions about whether an artist’s recognizable voice or identity can be treated as a protected commercial and biometric interest.

Music publishers are also pursuing claims against AI companies. Sony Music Publishing, Warner Chappell Music, and other publishers sued Anthropic over allegations that copyrighted musical compositions and lyrics were obtained through torrenting, scraping, and other unauthorized methods for use with Claude.

That case focuses more heavily on compositions and lyrics, while the new Suno lawsuit focuses on sound recordings. The difference matters because a sound recording and the underlying musical composition are separate copyrights, often controlled by different parties.

Labor groups are involved in the debate as well. The American Federation of Musicians has opposed dismissal motions filed by Universal and Warner in a case concerning alleged AI licensing of union-covered recordings. AFM argues that licensing recordings to AI companies for training and output generation may qualify as a “new use” under the union’s agreement, requiring additional compensation and information for musicians.

Together, these cases show that the AI music dispute is developing across several legal areas:

• Copyright in sound recordings
• Copyright in compositions and lyrics
• Rights of publicity
• Biometric privacy
• Collective bargaining agreements
• Licensing and compensation
• The legality of training data and model development

Radio talking points

• Universal and Sony are suing Suno for a second time, not simply repeating their first case.
• The new lawsuit targets the v6 model and the question of whether a new model can inherit legal problems from earlier systems.
• The phrase “the fruit of the same poisoned tree” summarizes the labels’ argument that later model improvements do not erase alleged earlier copying.
• The broader debate is moving beyond copyright to include artist identity, voiceprints, union agreements, and consent.
• None of the allegations in the new lawsuit have been proven in court.

What independent artists should understand

What “trained on” means in plain English

When an AI company says a model was “trained on” music, it generally means the system was exposed to recordings or other data during development so it could learn patterns.

That does not necessarily mean the model stores a complete song in a way a person can play back. Instead, the system may learn relationships involving rhythm, melody, arrangement, instrumentation, vocal qualities, genre, and production techniques.

The legal question is whether making and using copies of copyrighted material to develop that system requires permission. Courts may also examine how the material was obtained, whether the use is commercial, whether the model competes with the original works, and whether licensed alternatives are available.

For artists, the important point is that training data is different from an ordinary stream or listen. A recording used to train a model may become part of a technology company’s development process, which is why licensing and consent are central issues.

Why consent and opt-in language in your agreements matters

Artists should read agreements that address artificial intelligence, machine learning, voice cloning, digital replicas, training data, and derivative uses.

An opt-in clause generally requires clear permission before a recording, composition, name, likeness, or voice can be used for a particular purpose. An opt-out clause may allow use unless the artist takes steps to refuse it.

Those terms can have very different effects. Agreements should make clear:

• Which recordings or performances are covered
• Whether AI training is permitted
• Whether voice or likeness may be modeled
• How compensation is calculated
• Whether the artist can withdraw consent
• How long the permission lasts
• Whether the permission applies to future technologies

The Suno litigation shows why vague language can create disputes years after a recording is released.

Why registering and documenting your recordings protects you

Independent artists should keep organized records showing who created, performed, wrote, produced, and owns each recording.

Useful documentation can include split sheets, producer agreements, session musician agreements, work-for-hire documents, copyright registrations, metadata, delivery files, and dated project sessions.

Registration does not prevent infringement by itself. It can, however, help establish ownership and preserve legal options. Clean documentation also makes it easier to identify which rights belong to the artist, a label, a producer, a publisher, or another collaborator.

Artists should also save evidence of unauthorized uses, including screenshots, URLs, platform pages, prompts, generated outputs, and dates. Evidence can become difficult to recover if a page or platform feature changes.

SoundExchange and master recording royalties

Artists who own or control their master recordings may also collect eligible SoundExchange royalties when their music is played on qualifying non-interactive digital services.

SoundExchange royalties generally relate to digital performance royalties for sound recordings, not publishing royalties for the underlying composition. Eligible uses can include certain internet radio, satellite radio, and other non-interactive digital audio services.

Artists and sound recording owners typically need to register their information and maintain accurate recording metadata so performances can be matched correctly. SoundExchange does not replace collection systems for publishing royalties, mechanical royalties, sync fees, or on-demand streaming income.

The new Suno lawsuit is about AI training and copyright, not SoundExchange royalty collection, but both issues point to the same basic lesson: artists need to know what they own, what they have authorized, and how their recordings are being used.

What happens next

The second Universal and Sony lawsuit will now move through the federal court process. Suno will have an opportunity to respond to the allegations, and the court may address questions involving copyright ownership, training methods, licensing, discovery, damages, and fair use.

The result could influence how AI music companies build models, how labels license recordings, and how artists negotiate agreements involving their voices, identities, and catalogs.

For now, the case remains unresolved. The major issue is whether a newer AI model can be considered legally separate from earlier systems when the plaintiffs argue that it learned from their outputs, preferences, or capabilities.

Sources: Music Business Worldwide, the reported complaint, Reuters reporting on the Jason Isbell case, and SoundExchange.