AI Music Fraud Explodes: Sony Battles Deepfakes

Sony Music is significantly increasing its efforts to combat streaming fraud, particularly the growing threat of AI-generated songs mimicking popular artists. By the end of September, Sony had formally asked platforms to remove more than 260,000 tracks. This number nearly doubled from the 135,000 requests made by the end of March, indicating an accelerating problem.

The music group highlighted the scale of "generative AI deepfake tracks that imitate our artists' voices and likenesses without permission." This issue, affecting both current and historic artists, is seen as harmful to artists, misleading to fans, and detrimental to the ethical use of artificial intelligence.

Beyond impersonation, streaming services are facing a flood of AI-generated tracks created in bulk specifically for bots to play repeatedly. This practice generates royalties for owners of these tracks, thereby reducing the revenue available for genuine artists. Label executives estimate that such streaming fraud could be costing the music industry as much as $2.2 billion each year.

French streaming service Deezer reported this past summer that an average of 90,000 AI-generated songs are uploaded daily, now constituting over half of all new music on its platform. Furthermore, Deezer found that up to 85 percent of streams for fully AI-generated tracks last year were fraudulent.

This AI streaming fraud is rapidly becoming an epidemic, with organized fraudsters reportedly using platforms like Suno and DistroKid to distribute AI songs and commit fraud. Both Sony and Universal Music are pursuing legal action against AI music companies, alleging that they build models by copying recordings without proper licensing.

However, some AI music companies refute these claims, stating that the labels' assertions are fundamentally flawed in both factual and legal grounds. The battle for artists' rights and the integrity of music streaming continues to be an uphill struggle.