AI Song Generator Suno Leaked Source Code, Data Collection Revealed
A data leak from AI song generator platform Suno has revealed that the company collected millions of songs from various streaming services and other sources to train its AI models.
AI song generator platform Suno has been the subject of a significant data breach, which has exposed the company's extensive data collection practices for training its AI models. According to a hacking report, Suno downloaded millions of songs and lyrics from platforms including YouTube Music, Deezer, and Genius. Evidence of this data collection comes from leaked internal source code, published by hackers.
The platform reportedly used material gathered from services such as Pond5 and Jamendo. Specifically, over 11,000 hours of audio material were collected from YouTube Music, and more than 17,000 hours of lyrics from Genius. Suno had previously stated that it uses publicly available data, but the leak points to a more systematic and broader data acquisition strategy.
In addition to the source code, hackers also gained access to Suno's customer database. This database reportedly contained user email addresses, phone numbers, and partial credit card data stored via Stripe. However, Suno has decided not to notify its users about the security breach.
Suno has defended its decision, referring to it as a "limited security incident" and stating that the leaked source code was outdated. The company noted that no full credit card numbers were compromised, and therefore, it did not believe it had a legal obligation to issue an alert. The company now faces potential public backlash and scrutiny over its data handling and security practices.