SoundCloud’s AI Backpedal: What African Creators Need to Know About the Terms of Use Scandal
So, SoundCloud lit the internet on fire (again) over a 2024 change to their terms of use — one that made it seem like they might use indie artists’ tracks to train AI. Cue: panic, outrage, and a whole lot of “WTFs” from the artist community.
They’ve since backpedaled faster than a manager who forgot your split sheet. CEO Eliah Seton came through with an open letter clarifying that “SoundCloud has never used artist content to train AI models. Not for music creation. Not for large language models. Not for anything that tries to mimic or replace your work. Period.”
He added that SoundCloud isn’t dabbling in generative AI themselves — and they’re not letting third parties scrape artist content either. A “no AI” tag is now part of the platform’s toolkit to keep your work protected. Props for that, at least.
But here’s the tea: the wording in their 2024 terms was messy. It sounded way too broad, leaving artists (rightfully) suspicious. The new terms, updated May 14th, are supposed to clean that up — now clearly stating that SoundCloud won’t use your content to train generative AI models that could mimic your voice, music, or style.
Seton says if they ever consider AI for creators, it’ll be opt-in only, fully transparent, and with explicit consent. Translation? “No AI trickery unless you say yes.”
Still, not everyone’s vibing. Ed Newton-Rex of Fairly Trained called out the loophole: the new terms protect against mimicry, sure — but they still leave room for models trained on your work in sneakier ways. His clapback: “If they want to fix this, just say, ‘We won’t train AI on your music without your consent.’ Simple.”
Why This Matters for African Artists: AI is moving fast, and if platforms don’t keep it 100, your rights — your voice — could be compromised. At Downtown Music Publishing Africa, we’re watching this space like a hawk with a calculator. Our mission? To keep your music protected, your legacy intact, and your business handled.
Because your music isn’t training data. It’s intellectual property. It’s cultural capital. And it deserves more than a “sorry for the confusion” blog post.
