SoundCloud says it’s revising its terms after widespread backlash over a clause related to AI model training.

Earlier this year, SoundCloud quietly updated its usage policies, adding wording that many users interpreted as legal cover to allow the company to train AI on audio uploaded to its platform. SoundCloud was quick to assert that it wasn’t developing AI by using its users’ content, but the company’s PR statement didn’t allay fears that SoundCloud might do so in the future.

On Wednesday, SoundCloud CEO Eliah Seton published an open letter admitting that the wording of the changes to the company’s terms “was too broad and wasn’t clear enough.” The updates were intended to focus on other uses of AI internally at the company, Seton said — including recommendations and tools to help prevent fraud — but missed the mark.

SoundCloud has now revised its terms “to make it absolutely clear [that] SoundCloud will not use [user] content to train generative AI models that aim to replicate or synthesize [a] voice, music, or likeness,” said Seton.