SOCAN, the Canadian rights organisation representing more than 200,000 songwriters, composers and publishers, announced legal action against Suno on 2 September. Its stated case is that the platform "is producing and streaming outputs that replicate human-created musical works without consent or payment, and as a result has infringed the performing rights in musical works in SOCAN's repertoire."
A different claim from the American cases
The litigation against generative music platforms so far has been about inputs: whether copying recordings into a training corpus is infringement, and whether fair use covers it. SOCAN's pleaded cause is about outputs — that generated tracks reproduce protected works, and that producing and streaming them engages the performing right. That is why a performing rights organisation is the plaintiff rather than a label.
Why the distinction is strategic
An output claim does not require litigating what is in the training set, which is the most expensive and most contested part of the American cases. It requires showing that generated material replicates protected works and that the platform's distribution of it is a public performance. The evidentiary burden moves from a corpus nobody outside the company can inspect to material the platform publishes.
What the common framing gets wrong
Two things. This is being filed into the running narrative of "the music industry versus AI training," and it is not a training-data case. Reporting it as one erases the legal theory that makes it distinctive and testable on a different timetable. Second: an announced claim is an allegation, tested by nothing yet. Suno has contested the theories against it vigorously elsewhere, and nothing in this announcement has been ruled on.
What it opens
Performing rights organisations exist in nearly every jurisdiction and collectively license the public performance of essentially all commercial music. If the output theory succeeds anywhere, the exposure is not one company's training decisions but the ongoing distribution of every generated track — a licensing question rather than a one-time liability. That is a structurally different threat to a generative music business than a training-data judgment.