Sony removes 135,000 'deepfakes' of its artists' music - BBC

Sony Music has taken aggressive action against the surge of AI-generated content, requesting the removal of over 135,000 tracks that impersonated its artists on streaming platforms. These deepfakes, which utilize generative AI to clone the voices of high-profile musicians, are being used by fraudsters to skim money from streaming services by capitalizing on the popularity of genuine recording artists.

What happened

The scale of the issue is significant, with Sony identifying approximately 60,000 fraudulent tracks since March 2023 alone. The deepfakes target some of the company’s biggest names, including Beyoncé, Queen, Harry Styles, Bad Bunny, Miley Cyrus, and Mark Ronson. According to Dennis Kooker, president of Sony's global digital business, these attacks are often demand-driven, specifically targeting artists during new album release campaigns to hijack public interest and redirect potential revenue.

Beyond direct financial harm, the unauthorized content risks tarnishing artist reputations and creating confusion among fans who struggle to distinguish between human creativity and AI-generated counterfeits. The industry views this as a form of streaming manipulation, where AI is used to "supercharge" fraudulent activity by artificially boosting play counts to secure royalty payments. While Sony believes the 135,000 removed tracks represent only a portion of the total fraudulent content uploaded, the company and the International Federation of the Phonographic Industry (IFPI) are pushing for standardized identification and labeling tools. There is a growing industry consensus that streaming services must implement transparent detection systems, similar to those already utilized by platforms like Deezer, to protect the integrity of the music ecosystem and ensure that legitimate artists are not robbed of their earnings.