Suno has admitted in a court filing that it obtained audio data from YouTube for use as training data.
The admission appears in the AI music company’s answer to the first amended complaint from UMG Recordings, Capitol Records, and Sony Music Entertainment.
Suno also alleges that the labels have engaged in “anticompetitive activities that extend an unlawful monopoly” over music.
And it challenges whether they have standing to bring their YouTube “stream ripping” claim.
The 26-page document was filed on September 1 in the US District Court for the District of Massachusetts and reviewed by MBW. You can read it in full here.
“Suno admits that audio data was obtained from YouTube for use as training data using YT-DLP,” the filing states in response to paragraph 49 of the labels’ complaint.
That paragraph alleges Suno used two tools, YT-DL and YT-DLP, to “circumvent YouTube’s encryption and scrape copyrighted recordings from YouTube.”
Suno‘s response names only YT-DLP.
Suno‘s second affirmative defense argues that the labels’ claims are barred in whole or in part by copyright misuse and unclean hands – language it has pleaded, word for word, since its first answer in August 2024.
“On information and belief, Plaintiffs have engaged in anticompetitive activities that extend an unlawful monopoly over the production and commercialization of music,” reads Suno‘s second affirmative defense.
“Suno admits that audio data was obtained from YouTube for use as training data using YT-DLP.”
SUNO, IN ITS ANSWER TO THE FIRST AMENDED COMPLAINT
“To the extent there is copying of copyrightable expression, that copying constitutes fair use pursuant to 17 U.S.C. § 107,” reads Suno‘s first affirmative defense. “Suno’s AI tool uses a back-end technological process, invisible to the public, in the service of creating an ultimately non-infringing new product.
“This is quintessential fair use.”
Judge F. Dennis Saylor IV granted the labels leave to bring the stream-ripping claim on August 18, under the anti-circumvention provisions of the Digital Millennium Copyright Act, as MBW reported.
Suno had asked the court to throw the claim out in October 2025, arguing that the statute bars circumventing controls on access to a copyrighted work rather than controls on copying it.
Its lawyers called the claim “a gambit to try to evade application of the fair use doctrine to Suno‘s technology development process.”
Suno argues in the answer that the labels lack standing to bring the circumvention claim, saying “one or more” of them fail the test under Article III of the US Constitution.
It says the labels are not “person[s] injured” within the meaning of Section 1203(a), and therefore cannot maintain a civil action over any alleged circumvention.
Both defenses are limited to the circumvention claim, and do not touch the two copyright infringement claims the labels also bring.
Those two defenses are the only grounds on which Suno contests the court’s subject matter jurisdiction.
Suno told the labels in May 2025 that it had downloaded audio files from YouTube, and that it had used open-source tools such as YT-DL and YT-DLP, according to Saylor‘s order.
The September 1 answer restates that disclosure as a formal admission in Suno‘s own pleading.
The answer repeats at five further points that Suno “obtained audio data from YouTube” for use as training data.
Suno denies the remaining allegations, and says others contain legal conclusions requiring no response.
Udio, the rival AI music company sued alongside Suno in 2024, filed its own answer to Sony Music on April 29.
Udio used the same wording on fair use and copyright misuse, down to the phrase “quintessential fair use.”
The two AI companies share the same legal counsel: Latham & Watkins filed both answers, with Andrew M. Gass, Brittany N. Lovejoy, and Sarang V. Damle named on each.
Udio‘s filing carries a second firm, Quinn Emanuel Urquhart & Sullivan, that does not appear on Suno‘s.
Among its 13 affirmative defenses, Suno argues that any infringement was innocent, which it says would allow the court to reduce statutory damages to as little as $200 per work.
The labels are seeking up to $150,000 per work across the 560 recordings in suit, after Saylor denied their bid to add 61,026 more without prejudice.
Suno also argues that some or all of the material the labels claim copyright over is in the public domain, that the registrations are invalid, and that they have suffered no provable injury.
Elsewhere, Suno “admits that Plaintiffs do not appear to be alleging that the outputs” of its model infringe the recordings in suit.
Suno denies that similarities between those outputs and the labels’ recordings “betray that the models were trained on the Copyrighted Recordings.”
On the technology at the center of the claim, Suno says it “presently lacks knowledge or information sufficient to form a belief” about YouTube‘s “intent in designing any particular technology.”
On the scale of its training, Suno admits that its model was built by showing the program “tens of millions of instances of different kinds of recordings gathered from publicly available sources.”
In its first answer, filed in August 2024, Suno said its training data included “essentially all music files of reasonable quality that are accessible on the open Internet.”
That filing named no source and pleaded 11 affirmative defenses, none about standing.
The new answer admits that over 12 million users have generated music files using Suno‘s product, and that its Pro and Premier plans cost $8 and $24 per month.
Warner Music Group settled its case against Suno in November 2025 and struck a licensing deal, leaving UMG, Capitol and Sony Music as the remaining plaintiffs.
Fact discovery closes on September 30, with both sides expected to move for summary judgment on the fair use question.
Suno asked the court to enter judgment in its favor, dismiss all claims with prejudice, and award its attorneys’ fees and costs.
Away from the litigation, the platform said on August 10 that it would “introduce a new generation of Suno models, developed in partnership with the music industry,” and that “all prior models will be retired” when those launch.
The Suno blog added that the models are “better than anything we’ve ever released” on “every metric we’ve ever measured,” and would “deliver higher fidelity audio.”
Suno‘s statement gave no launch date and no version number.
A video posted to Suno‘s Instagram account in early September carries the name V6 and the words “New models. Coming soon.”
Asked by an Instagram user about stems, the Suno account said “v6 has reduced bleeding and better timing on stems, plus higher fidelity overall with fewer artifacts.”
Asked about musical key, Suno said “v6 will have stronger adherence to musical key so you can call it out right in your prompt.”
MBW reported in March 2026 that no new model had launched, after Suno pledged on settling with Warner Music Group that its current models would be deprecated when licensed models arrived in 2026.
Suno signed a licensing deal with BMG on August 12, covering the company’s recorded music and publishing repertoire.
BMG says it is a participant, alongside Warner Music Group, in the creation of Suno‘s upcoming models.
BMG was not a plaintiff in the 2024 lawsuit, and says the deal settles Suno‘s prior use of its works.
Starting September 3, Suno is capping monthly downloads at 20 for Pro subscribers, 60 for Premier, and 7 in total for free users.
Suno‘s new Terms of Service, published on August 10, prohibit obtaining an output by any means other than its own download channel, and give “recording or stream ripping” as examples.Music Business Worldwide
Suno, an AI music company, has recently admitted in a court filing that it obtained audio data from YouTube to use as training data for its AI models. This admission comes in response to a first amended complaint from major music labels, including UMG Recordings, Capitol Records, and Sony Music Entertainment, which allege that Suno engaged in “stream ripping” by using software tools to circumvent YouTube’s encryption to download copyrighted recordings.
The court filing, which spans 26 pages and was submitted to the US District Court for the District of Massachusetts on September 1, outlines Suno’s defenses against the claims. Notably, Suno’s legal team acknowledges the use of YT-DLP, a tool for downloading videos from YouTube, as part of their data collection process. However, they contest the labels’ claims, arguing that the labels have engaged in “anticompetitive activities” that amount to an unlawful monopoly over music distribution.
In its response to the labels’ allegations, Suno asserts that any potential infringement falls under the fair use doctrine, as defined by 17 U.S.C. § 107. Suno claims that its AI technology operates in a way that ultimately produces non-infringing new products, constituting “quintessential fair use.” The company’s first defense highlights that any copying of copyrightable material is justified under this legal framework.
Moreover, Suno’s legal team argues that the music labels lack standing to pursue their claims. They assert that at least one of the plaintiffs does not meet the necessary criteria as defined under Article III of the US Constitution, which requires that a plaintiff demonstrates they have suffered an injury to maintain a civil action. This argument is particularly aimed at the circumvention claim related to the Digital Millennium Copyright Act (DMCA), which prohibits circumventing technological protection measures around copyrighted works.
Suno’s position is that the DMCA’s anti-circumvention provisions apply to protecting access controls rather than copying controls, and they contend that the labels’ claims attempt to circumvent the fair use doctrine applicable to Suno’s development process. The company maintains that its model was trained on « tens of millions of instances of different kinds of recordings gathered from publicly available sources, » and that it did not infringe on the copyrights of the labels’ specific recordings.
In addition to its legal defenses, Suno’s filing also outlines various affirmative defenses, including claims of innocent infringement, which could allow for reduced statutory damages if the court finds any infringement occurred. The labels are seeking substantial damages of up to $150,000 per work across 560 recordings, and Suno is pushing back by asserting that many of the materials claimed by the labels are either in the public domain or that the registrations might be invalid.
The legal landscape is further complicated by the fact that Udio, another AI music company involved in similar litigation, has filed responses with comparable legal arguments. Both companies share legal representation from the law firm Latham & Watkins, which indicates a coordinated defense strategy.
Suno’s history with the litigation has included a previous settlement with Warner Music Group in November 2025, which resulted in a licensing deal. This left UMG, Capitol, and Sony Music as the remaining plaintiffs in the ongoing case. The court is expected to move toward summary judgment regarding fair use, with fact discovery closing on September 30, 2026.
In the midst of these legal challenges, Suno announced plans to launch a new generation of AI models, developed in collaboration with the music industry. The company has stated that these new models will provide significant improvements in audio fidelity and performance metrics compared to previous versions. However, a specific launch date has not been disclosed, and the company has been cautious about releasing new technology amid ongoing litigation.
Additionally, Suno has implemented changes to its service offerings, including capping monthly downloads for users on different subscription plans and updating its Terms of Service to prohibit certain methods of obtaining outputs. The platform has emphasized a shift away from previous models to align with its new licensing agreements and the needs of the music industry.
As the case progresses, both Suno and the music labels are preparing for potential summary judgment motions regarding fair use, which could have significant implications for the future of AI-generated music and the legal boundaries of copyright in the digital age.
