Open letter: SoundCloud CEO Eliah Seton says a 2024 TOS update was not intended to allow AI training on user content; SoundCloud revised the wording on May 14
- Click to email a link to a friend (Opens in new window) Email — Music Ally NEXT 2025: Tickets on sale now!
Context & Ripple Effects
The clarification follows reporting that SoundCloud’s policy language had expressly permitted uploaded audio to be used to “inform, train, develop or serve as input” to AI. The company is now recasting that language through a revised AI-use policy.
The dispute lands in a music market already sensitive to training-data rights: Sony Music’s warnings to AI companies and streaming platforms underscored rights holders’ resistance to unlicensed model training. As SoundCloud adds more community-facing product features, creator confidence becomes more central to the platform’s operating model.
First-order effects
- SoundCloud’s May 14 wording revision reduces the immediate ambiguity around whether uploads can be used for AI training, while putting Seton’s interpretation of the 2024 update on the record.
- Artists and other uploaders gain a clearer basis for assessing how their audio may be used; SoundCloud must align any AI-related use with the revised terms and its public explanation.
Second-order effects
- The episode raises the cost of broad, opaque AI clauses for audio platforms: creators, labels, and partners can scrutinize terms changes sooner and demand clearer consent boundaries.
- Music services pursuing AI features face added pressure to distinguish licensed inputs from platform-uploaded material, particularly after calls for services to downweight AI music trained without licensed data.
Third-order effects
- If platforms continue to treat user uploads as potential model inputs, terms of service will become a more important governance and trust layer—not merely a legal backstop—for creator platforms.
- The likely structural direction is toward more explicit disclosure and permission design around training data, though this case alone does not establish how consistently platforms will adopt it.
The trend: Creator platforms are being pushed to make AI-training permissions more explicit as user-generated content becomes valuable model input.