Hacked Data Reveals Suno Built Its AI Music Engine on Scraped YouTube and Deezer Content

A data breach at AI music startup Suno has pulled back the curtain on one of the music industry's most closely guarded secrets: exactly where these AI generators get their training material. According to a report by 404 Media, internal data obtained through a hacking incident reveals that Suno systematically scraped millions of songs and lyrics from major online platforms, including YouTube Music, Deezer, and lyrics database Genius. The company had previously declined to publicly disclose the contents of its training datasets or the methods used to acquire them, making this leak an unusually candid look at its data pipeline.

The timing of this exposure is particularly consequential for Suno. The Recording Industry Association of America, which represents major labels and artists, has already filed a lawsuit against the company alleging widespread use of copyrighted material without authorization or compensation. In connection with that legal action, Suno reportedly admitted to certain practices — though the full scope of those admissions has not been entirely public until now. The newly leaked data appears to corroborate and potentially expand upon the allegations that have been leveled against the startup.

Suno is one of the most prominent players in the booming AI music generation space, allowing users to create full songs from text prompts within seconds. Its technology has attracted significant investor interest and a large user base, but the legal questions surrounding its training data have shadowed its rise. The company has maintained a stance of opacity regarding its data sourcing, a position that is now increasingly difficult to sustain in light of both the litigation and the breach.

The broader implications stretch well beyond Suno itself. Numerous AI companies across music, image, and text generation have faced similar accusations of training on scraped, copyrighted content without licensing agreements. Courts and regulators in multiple countries are still grappling with whether such practices constitute fair use or outright infringement. The Suno leak may serve as a pivotal data point in those ongoing legal and legislative debates, giving plaintiffs and policymakers concrete evidence of how these systems are actually built.

For artists and rights holders, the revelation is both validating and alarming. Many musicians have long suspected that their recorded work was being used to train commercial AI tools without their consent or compensation. Suno's case may accelerate industry-wide calls for mandatory disclosure of AI training datasets, licensing frameworks that include creators, or even legislative mandates requiring transparency from AI developers about the origins of their training data.