AI war on streaming
Global recorded music revenue hit $31.7B in 2025 with streaming accounting for ~70% (over $22B), and platforms like Deezer received more than 60,000 fully AI‑generated tracks in January 2026 — critics say play counts are being inflated and creators underpaid (aceshowbiz.com) (interaksyon.philstar.com). Right now Sony Music has targeted 135,000+ deepfakes for removal and BMG has sued Anthropic claiming its Claude model was trained on copyrighted lyrics — the legal and rights fight over AI music is active and escalating (musicbusinessworldwide.com) (computerworld.com).
IFPI’s Global Music Report says there are now 837 million paid streaming subscribers worldwide and that paid-subscription receipts grew 8.8% in 2025, with Latin America posting the fastest regional growth at 17.1% year‑on‑year. (ifpi.org) Deezer reports it detected and tagged more than 13.4 million AI‑generated tracks over 2025 and says up to 85% of streams on AI‑generated music were flagged as fraudulent and removed from the royalty pool. (musicbusinessworldwide.com) The Paris platform is now offering its AI‑detection technology commercially after testing with partners including French collecting society SACEM, and says its system can identify fully synthetic output from models such as Suno and Udio and exclude flagged tracks from algorithmic recommendations. (musicbusinessworldwide.com) At the IFPI report launch Sony executives named high‑profile roster acts — including Beyoncé, Queen, Harry Styles, Bad Bunny, Miley Cyrus and Mark Ronson — as targets of fraudulent AI uploads and told the BBC such fake tracks can “damage a release campaign or tarnish the reputation of an artist.” (musicbusinessworldwide.com) Sony also told industry bodies it had identified roughly 60,000 falsely attributed tracks since March 2025 and warned the volume is likely only a fraction of total uploads, while urging platforms and aggregators to act on streaming manipulation. (musicbusinessworldwide.com) BMG filed a federal complaint against Anthropic on March 17 alleging 493 examples of copyright infringement — naming lyrics and works tied to artists including Bruno Mars, Ariana Grande and the Rolling Stones — and saying its models reproduce substantial song text after being trained on unlicensed material. (money.usnews.com) Anthropic previously agreed to a roughly $1.5 billion settlement with a class of book authors over similar claims about training data, a development rights holders cite as precedent while multiple publishers continue separate litigation over AI training practices.
