Suno v6 Faces New Copyright Battle as Sony Music and UMG Challenge AI Training Methods
The legal battle between the traditional music industry and generative AI has entered another major phase, with Suno v6 now at the center of a new copyright dispute involving Sony Music Entertainment and Universal Music Group (UMG).
Suno launched its v6 generation of AI music models on September 9, 2026, describing the release as a new chapter built with music-industry partners including Warner Music Group, BMG and Believe. The company positioned v6 as a more sophisticated and industry-connected generation of AI music technology.
But just days later, Sony Music and UMG filed another lawsuit against Suno. The labels argue that creating a new AI model from the outputs and user interactions associated with previous models does not erase the copyright issues they say originated with those earlier systems. Their complaint characterizes v6 as being connected to the allegedly unauthorized use of copyrighted recordings in previous Suno models.
Suno strongly disputes the labels' interpretation and says v6 was developed using licensed partner content, community interactions and the accumulated technical knowledge of its team.
The dispute could become an important test of how copyright law applies not only to the original training of AI models, but also to the way new generations of models learn from earlier AI systems.
Suno v6 Was Supposed to Represent a New Beginning
When Suno introduced v6 on September 9, the company presented it as a substantial technological upgrade rather than simply another incremental model release.
Suno says v6 was developed alongside industry partners including Warner Music Group, BMG and Believe. The company described three versions: the flagship v6 model, the more experimental v6-wild and the faster v6-mini.
The new generation also introduced a much broader set of creative tools.
Users can edit sections of songs with natural-language instructions, create mashups, sample and isolate musical elements, change individual lyrics and use text, audio, images and video as creative inputs. Suno also said v6 understands musical structure, instrumentation, vocals, mood and other elements with greater precision.
The company had also been moving toward a more licensed approach to AI music.
Before v6 arrived, Suno had settled with Warner Music Group and subsequently announced relationships with BMG and Believe. Those developments were part of a broader effort to create an AI music ecosystem in which rights holders could participate in licensing and future revenue opportunities.
For Suno, therefore, v6 represented more than a technical upgrade. It was also part of the company's attempt to establish a different relationship with the music industry.
Sony Music and UMG Say v6 Still Has a Copyright Problem
Sony Music and UMG have a very different interpretation of how v6 was developed.
In their September 18 lawsuit filed in federal court in Boston, the labels accuse Suno of copying 60,202 sound recordings without authorization and using them in the development of its AI technology. The complaint also raises allegations involving the downloading of music from YouTube.
The most important argument surrounding v6 concerns the relationship between old and new models.
Sony and UMG argue that even if v6 was not directly trained on the labels' original recordings, the model could still have benefited from previous Suno models that allegedly were trained on copyrighted material.
The labels describe this process as “model laundering.” Their argument is essentially that creating a new model from the outputs or learned behavior of an earlier model does not necessarily remove the legal consequences associated with the earlier model's training.
That distinction could become extremely important for the entire AI industry.
Why Training One AI Model on Another Matters
AI models are not always developed as completely isolated systems.
Companies can use different techniques to improve new models, including feedback from users, preference data, generated outputs, model distillation and other forms of information derived from earlier systems.
From a technology perspective, this can be an efficient way of improving performance.
From a copyright perspective, however, the question becomes more complicated.
If an earlier model was allegedly trained using copyrighted recordings without permission, can the developer use outputs, preferences or learned information from that model to build a new system?
Sony and UMG argue that the answer should not automatically be yes.
Their complaint alleges that Suno's newer models remain connected to the earlier systems because user interactions and generated outputs were incorporated into the development process. The labels contend that these outputs themselves were influenced by earlier models.
The legal question is therefore broader than whether a copyrighted recording can be found directly inside v6's training dataset.
It could involve how knowledge moves from one AI model to another.
Suno Says v6 Uses Licensed Content and Community Data
Suno has rejected the labels' allegations.
Following the new lawsuit, the company said v6 was trained using content licensed from its partners, interactions including creations and preference signals from its community, and accumulated learnings from its team.
That statement is particularly significant because it provides more detail about what Suno means when it discusses “user data.”
At the launch of v6, Suno Chief Product Officer Jack Brody said the model was trained from scratch and included licensed partner data, user data and the company's accumulated technical learnings.
Suno later clarified that its description of community interactions included user creations as well as preference signals.
The distinction between those categories is likely to receive significant attention as the case develops.
The labels argue that user-generated outputs and preference signals can still be connected to earlier models that allegedly incorporated copyrighted recordings.
Suno, meanwhile, maintains that its current approach represents a legitimate path toward building AI music products in collaboration with the music industry.
Suno's YouTube Training Admission Adds Another Layer
The v6 dispute cannot be separated entirely from the earlier Suno copyright litigation.
In September, Suno acknowledged in a court filing that it obtained audio data from YouTube for use as training data using YT-DLP. That admission relates to the original litigation brought by the major labels and is separate from Suno's description of the data used to build v6.
Suno continues to argue that its use of copyrighted material for AI training can qualify as fair use.
The company has also challenged parts of the labels' claims and has raised defenses including copyright misuse and arguments concerning the labels' standing on the separate stream-ripping allegations.
This means the legal fight is not simply about v6.
There are multiple connected questions involving the original training process, subsequent models, YouTube data, fair use and whether information derived from previous AI models can legally be incorporated into new generations.
The Original Suno Lawsuit Is Still Important
Sony and UMG's latest case follows the original lawsuit filed against Suno in 2024.
That earlier litigation initially involved 560 recordings. Warner Music Group was also part of the original dispute before reaching a settlement with Suno in November 2025 and entering into a licensing agreement. Warner's entities were subsequently dismissed from that case.
Sony and UMG remain involved in the continuing litigation.
The labels have also sought to expand the scope of the dispute. In August, a federal judge allowed Sony and UMG to add a claim accusing Suno of circumventing YouTube's anti-downloading technology, while denying their request at that stage to add more than 61,000 recordings to the existing lawsuit.
The judge indicated that resolving the central fair-use question was important to moving the case forward.
That makes the Suno litigation one of the most closely watched legal battles in the rapidly developing AI music sector.
Why This Case Matters for AI Music
The implications extend well beyond Suno.
If courts ultimately determine that AI companies can be liable for using copyrighted recordings to train models without permission, that could strengthen the argument for licensing agreements across the generative music sector.
On the other hand, if courts establish broader protection for certain forms of AI training under fair-use principles, developers could have more room to build models using large collections of copyrighted material.
The v6 issue adds another potential dimension: model-to-model learning.
AI companies may increasingly build new systems using information generated by previous systems. If courts determine that the legal status of the original model can affect the legal status of subsequent models, developers could face additional obligations when upgrading their technology.
That could influence how companies design datasets, document model development and retain information from previous systems.
The Music Industry Is Moving Toward Licensing
One notable development is that the music industry is not taking a single approach to AI.
Warner Music Group has reached an agreement with Suno. BMG and Believe have also partnered with the company, while Suno says these relationships helped form part of the foundation for v6.
Suno has also discussed future opt-in products where individual artists could participate and potentially receive compensation.
The company's stated goal is to create products where AI and the traditional music business can work together rather than remain entirely opposed.
At the same time, Sony and UMG are continuing to pursue litigation.
That contrast illustrates one of the central questions facing the industry: what should licensed AI music look like, and how should creators and rights holders be compensated?
What Happens Next for Suno v6?
The legal process is still developing, so there is no final court ruling establishing whether Sony and UMG's allegations about v6 are correct.
The labels have presented their theory of the case, while Suno disputes the claims and maintains that its v6 technology was developed through a combination of licensed content, community interactions and internal technical development.
The original Suno litigation is also moving toward arguments over whether AI training on copyrighted recordings without a license constitutes fair use.
As discovery and court proceedings continue, the music industry will be watching for answers to several fundamental questions.
Can an AI company train a new model using outputs from an older model? Does the legal status of the older model affect the newer one? How should user-generated AI music be treated when it becomes training data? And where should the line be drawn between copyrighted expression and general musical knowledge?
Those questions could shape AI music development for years.
Suno v6 Could Become a Defining AI Music Copyright Case
Suno's v6 launch was designed to demonstrate how AI music technology could evolve alongside the traditional music industry.
Instead, it has quickly become part of a major legal debate over how AI models inherit information from their predecessors.
Sony Music and UMG argue that a new model cannot simply escape the legal consequences of an earlier model by being rebuilt using generated outputs or user interactions. Suno argues that v6 represents a new generation built with licensed industry partners, community feedback and its own technological development.
At this stage, those are competing legal positions rather than established findings.
What ultimately matters will be the evidence presented in court and how judges interpret copyright law, fair use, AI training and the relationship between successive generations of machine-learning models.
For musicians, labels, AI developers and music fans, the outcome could help define the rules of one of the fastest-changing areas of the modern music business.
