The Jerusalem PostSuspect arrested in fatal shooting of 30-year-old man in Jaffa as police probe motive of incidentESPN DeportesBoston Celtics: resumen de temporada baja y previa 2026-27וואלהצה"ל: חוסל ראש חולייה מארגון הטרור גא"פ שפשט לכיסופים ב-7 באוקטוברRTP DesportoBenfica aponta ao tricampeonato de futsalInquirerDy thanks Marcos for directing VAT removal on system loss chargeDaily MaverickCULTURAL PHENOMENON: Star Trek at 60 and the optimistic future it still dares us to imagineESPNLeagues, unions urge action against escalating threats from bettorsDeadlineAmazon Prime Video Debuts $30-A-Month Bundle With AMC+, BritBox, MGM+, PBS Masterpiece & StarzABC NewsMeasles-related deaths in Pennsylvania rise to 4, health officials sayVarietyQuentin Tarantino to Publish ‘Cliff Booth’ Novel Before Brad Pitt and David Fincher’s Movie Streams on NetflixCBS Sports2026 Week 2 NFL odds, start times, betting lines, spreads: Get Week 2 NFL picks, predictions for every gameXatakaVivió hasta los 103 años, actuó en más de 90 películas y fue nominado a tres Óscar. Hoy su hijo es toda una leyenda de Hollywood
The Daily Newsstand · Free, Always
Tuesday, September 15, 2026

Suno admits using music from YouTube to train AI models

Translate

Suno has admitted to using music from YouTube to train its AI models.

The AI music creation firm is currently battling a lawsuit brought forth by Universal Music Group (UMG), Sony Music, and Capitol Records concerning its allegedly unauthorised use of licensed music for training its models.

As reported by Music Business Worldwide, the latest court documents were filed in the US District Court for the District of Massachusetts on September 1, and in them, Suno admits to having used “audio data obtained from YouTube” as training data.

The labels allege that Suno obtained music from YouTube by using software designed to download audio from online videos, and an amended complaint names tools including YT-DL and YT-DLP, which it’s alleged that Suno used to bypass YouTube’s technical protections and extract copyrighted recordings.

As noted by Music Business Worldwide, lawyers on behalf of Suno have argued that the three plaintiffs are “engaging in anticompetitive activities that extend an unlawful monopoly over the production and commercialisation of music,” and that the company’s use of copyrighted music constitutes “fair use” due to its AI models creating an “ultimately non-infringing new product”.

Last year, Warner and fellow major music labels Sony Music Entertainment and Universal Music Group filed lawsuits against Suno and another AI company Udio for copyright violations of “an almost unimaginable scale”.

However, Warner later signed a new licensing deal with Suno, settling the lawsuit in a deal reported at the time to be designed to introduce a licensed model into Suno’s offerings, making it such that users can pay to download songs that were made using AI.

Pitchfork reported that the deal ensures artists and songwriters who opt into AI deals with Warner and Suno will be compensated fairly, and will retain “full control” of their music, likeness and other copyright details, while also overseeing how AI uses their likeness.

At the centre of the case seems to be one of the biggest issues surrounding the creation of AI music – namely, where the material used to train it comes from, and how artists can be properly compensated.

Earlier this month, this saw artists including Jason Isbell file a lawsuit against Suno, alleging that the software allows users to copy artists’ identity and musical style without consent.

Last month, Suno lost a copyright infringement lawsuit to German licensing agency GEMA, after the latter argued that Suno was using songs in its repertoire – including Boney M’s ‘Daddy Cool’, Lou Bega’s ‘Mambo No. 5’, and Alphaville’s ‘Forever Young’ – to train its AI models, without obtaining any licenses or paying the songwriters.

That ruling also meant that other AI companies must now pay to license any music from GEMA’s repertoire going forward.

View the original on NME

KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.