Once thought to be the tool that worked while humans create art, AI now composes the very songs we thought were ours to produce. AI-generated tracks now make up roughly a fifth to a quarter of new uploads to major streaming platforms, and most trace back to one company: Suno, a Cambridge startup built by four former Kensho researchers. How it works is by turning a short text prompt into a finished song with vocals, structure, and mixing, and scarily, there is no real competitor to compete with. Udio came closest but got pulled into the same legal fight and shifted its strategy toward enterprise licensing. Meta, Google, and Stability AI built comparable technology but never shipped it as a consumer product in fear of the same legal trouble. While other AI companies focus on chatbots, this vacuum in technology let Suno set the terms for how AI music entered the market, with no competitive check on pace or scale.
The copyright fight triggered by that scale has moved since the RIAA first sued Suno and Udio in June 2024. Warner Music settled with Suno in November 2025, dropping its suit and signing a licensing partnership, while Universal Music Group and Sony Music continue to litigate against Suno in Massachusetts federal court. A summary judgment hearing on fair use is scheduled for July 2026 before Judge Denise Casper, and the ruling could set the legal standard for whether training AI models on copyrighted recordings counts as fair use industry-wide. On the Udio side, UMG settled and secured a per-generation royalty between $0.002 and $0.005, a number that shows what labels think AI-generated output is worth per stream. Sony has not settled with either company, and musicians' union AFM has already sued Universal and Warner, alleging settlement money is being kept from session musicians rather than passed down.
The lawsuits will decide who gets paid. They won't decide what gets lost. Streaming platforms pay from a fixed royalty pool, so every stream captured by a machine-generated track is a stream taken from a human one, either intentionally or not. Discovery algorithms can't tell a six-month labor of love from a forty-five-second prompt, so both compete for the same slot. The deeper cost is generational when fewer people will spend years learning production and performance if a prompt produces something "good enough" for little to no cost. Music has survived the phonograph, radio, and streaming because a human being was still making it for another human being to hear. Suno is the first technology in that history that can remove the human factor from both ends or evolve the traditional music production into one distinguishable enough.
Sources: Reuters, The Guardian, LA Times
Photos: Unsplash
Written by: Ariff Azraei Bin Mohammed Kamal
