Suno hack reveals its AI songs are based on vast amounts of ripped music (www.techradar.com)

A recent hack of AI music generator Suno has unveiled that the platform has utilized vast amounts of copyrighted music from sources like YouTube Music and Deezer without permission. The hacker, known as ellie.191, accessed Suno's source code and confirmed that over 2 million tracks and more than 12,000 hours of music were unlawfully ripped for training its AI models. This includes tactics to extract acapella versions for vocal training and a focus on podcasts as well. This revelation is significant for the AI/ML community as it highlights ongoing ethical and legal concerns surrounding the use of copyrighted content for training AI systems. Suno is already embroiled in multiple lawsuits regarding these practices, with serious implications for the industry's future engagement with creative works. The case raises critical questions about fair use in the context of AI training, reflecting broader tensions across various creative fields, from writing to visual arts, where the balance between innovation and copyright protection remains contentious.