Beyoncé, Queen, and Harry Styles deepfake (Mar 2026)

In March 2026, Sony Music disclosed that it had requested takedowns of more than 135,000 generative-AI deepfakes impersonating its artists — fraudulent tracks and voice clones falsely attributed to acts including Beyoncé, Queen, and Harry Styles. The figure, revealed at the launch of the IFPI's Global Music Report on March 18, 2026, was nearly double the 75,000 takedowns the label had reported about a year earlier, making it one of the clearest public measurements of how fast AI impersonation is scaling in the music industry.

What happened

According to Music Business Worldwide, Sony Music identified over 135,000 fraudulent AI-generated tracks impersonating its roster, with roughly 60,000 flagged since March 2025 alone. The fakes circulated on major streaming and social platforms, where they competed directly with legitimate releases for listener attention and streaming revenue.

Dennis Kooker, Sony Music Entertainment's president of global digital business, described deepfakes as a demand-driven problem: the fraudulent uploads cluster around moments when an artist is actively promoting new music, exploiting search interest during promotional cycles. Sony argued the fakes cause direct commercial harm — siphoning streams from real releases, potentially damaging carefully planned release campaigns, and tarnishing artists' reputations when listeners mistake low-quality clones for official work.

The disclosure came amid policy battles over AI and copyright. Sony had raised the takedown numbers in its submission to the UK government's consultation on AI and copyright law, pushing back on proposals to loosen protections and advocating for transparency obligations and better detection of AI-generated content at the point of upload.

Why this incident matters

Most deepfake incidents involve a single victim and a single piece of media. This one revealed industrial scale: one label, 135,000 fakes, and a caseload that nearly doubled in a year. Voice cloning tools have made it trivial to publish a plausible "new" Beyoncé or Harry Styles track, and takedown-by-takedown enforcement is a whack-a-mole response to automated generation. The episode reframed music deepfakes from a novelty problem into a systemic streaming-fraud and artist-rights problem that labels, platforms, and legislators are still working out how to police.

Where detection fits

Enforcement at this scale cannot rely on human listeners. An AI voice detector can screen suspect tracks for the synthesis artifacts left by voice cloning, helping platforms and rights holders identify impersonations at upload rather than after a fake has already drawn streams during an artist's release window.