When an AI model spits out a song that sounds suspiciously like something a real person wrote, who gets paid? SOCAN — the organization that collects royalties for more than 200,000 Canadian songwriters, composers and music publishers — thinks it finally has the start of an answer.
On Tuesday, SOCAN announced a collaboration with Musical AI, a Canadian rights-management startup that builds consent and attribution infrastructure for generative music. The deal names Musical AI an “approved technology partner” and sets out to explore how its tools can trace an AI-generated track back to the human work that shaped it — and route credit and compensation accordingly.
The pitch rests on two principles the companies keep returning to: creators should decide whether their catalog trains AI at all, strictly on an opt-in basis, and they should be paid when their work influences a machine-made output. That’s a pointed stance in an industry where the default has too often been to scrape first and apologize later.
How the attribution actually works
Musical AI’s technology analyzes a generated output and estimates how much specific source material influenced it, down to percentages, separating the contribution of a sound recording from that of the underlying composition. The result looks a lot like a traditional “split sheet” — the document co-writers use to divide songwriting credit — except it’s generated after the fact for an AI output. In theory, that hands a performing rights organization like SOCAN something it has never had: a paper trail for royalties on AI music.
“SOCAN’s role is to protect the rights and livelihoods of songwriters, composers and music publishers. This collaboration with Musical AI establishes a foundation for attribution and compensation, to ensure that music creators are paid when their music is used by AI.”
Jennifer Brown, CEO, SOCAN
There’s a national-pride angle, too. Both companies are Canadian, and Evan Solomon — the country’s first Minister of Artificial Intelligence and Digital Innovation — publicly backed the tie-up as “responsible Canadian innovation.” SOCAN is openly framing the work as a template other PROs around the world might copy.
For now, though, this is a framework, not a payout. The announcement leans heavily on words like “explore,” “guidelines” and “in the future,” and it says nothing about how much money might change hands, which AI platforms would agree to be measured, or how disputes over those influence percentages get settled. Attribution only matters if the companies generating AI music actually run their outputs through it — and plenty have shown little appetite for being audited. Still, as a marker of where the rights business is heading, it’s a meaningful one.
