Sony Music says it had asked digital platforms to remove more than 260,000 AI-generated deepfake tracks by the end of September, up from about 135,000 at the end of March. The figures, reported by the Financial Times and specialist music press on 5 October, concern tracks that imitate artists’ voices or likenesses without their permission, including names such as Harry Styles, Adele and Michael Jackson. Sony described the problem as occurring “at scale” and said it expects platforms both to remove infringing material quickly and to prevent repeated reuploads. Sony’s total is a cumulative company count of tracks flagged for removal, not confirmed takedowns; separately, Deezer reported receiving nearly 90,000 fully AI-generated tracks per day in June 2026, a daily upload figure for one platform that includes synthetic music whether or not it impersonates an identifiable artist. The increase nevertheless provides a concrete indicator of how rapidly identity-based abuse is growing alongside mass-generation tools and streaming-fraud schemes.
IQMA COMMENTARY: WHAT IT MEANS FOR MUSICIANS
For musicians, identity protection is becoming an operational infrastructure problem rather than an occasional celebrity dispute. Takedowns after release are necessary but reactive: once a convincing fake circulates, audience confusion and reputational damage may already have occurred. Platforms would need pre-release identity checks, persistent provenance, rapid artist-controlled reporting, and penalties for repeat uploaders. While this creates a heavy administrative workload that platforms want to avoid, how can they do so? AI disclosure labels help only when bad actors cooperate. The stronger standard is consent: a recognisable voice or persona should not become a freely reusable production input merely because a model can reproduce it.