SoundCloud Revamps Terms of Service Amid AI Policy Controversy

SoundCloud has released an updated version of its terms of service just days following the music platform’s change.
caught heat from musicians
And supporters following an earlier revision of their guidelines on artificial intelligence training.

On Wednesday, SoundCloud CEO Elijah Seton released an open letter stating that the platform “has never utilized artists’ content for training AI models.” This reaffirms a message previously provided by a SoundCloud spokesperson.
The Hollywood Reporter
last week.

“Definitely not for music creation. Definitely not for large language models. And certainly not for anything aiming to imitate or take over your job. End of story,” Seton stated. “We do not develop generative AI tools, nor do we permit third parties to access or utilize artist content from SoundCloud to train such systems.”

The controversy arose following an update to SoundCloud’s terms of service in February 2024, which states that users “specifically consent to their content being utilized to educate, train, enhance, or act as input for artificial intelligence or machine learning technologies and services, all in support of delivering these services.”

Last week, that update began circulating and sparked anger among musicians who were concerned about their content being utilized to train generative AI systems. In his letter, Seton stated that similar to SoundCloud’s previous announcement from the prior week, the company had revised its terms of service “to clearly explain our potential internal usage of AI to enhance the platform for both artists and audiences.” He mentioned functionalities such as enhanced search capabilities, better curation of playlists, and more tailored content suggestions as examples.

Seton stated on Wednesday that the wording in that update was “too vague and insufficiently clear.”

“It created confusion, and that’s on us,” Seton wrote. “That’s why we’re fixing it.”

According to the letter, SoundCloud is updating their terms of service. They state, “We will refrain from using Your Content to develop generative AI models designed to mimic or reproduce your voice, music, or persona without obtaining your explicit permission via an affirmative opt-in process.” This new policy will replace the previous wording.

Seton said that “if there is an opportunity to use generative AI for the benefit of our human artists, we may make this opportunity available to our human artists with their explicit consent, via an opt-in mechanism.”

Ours is a clear stance: AI ought to assist musicians rather than take their place. We’ll keep this perspective as we move forward with any implementation of such technologies on SoundCloud,” stated Seton. “The musical world is evolving with AI being an integral component. This presents fresh possibilities along with significant hurdles. Hence, our strategy will consistently prioritize one guiding rule: putting artists first.

SoundCloud’s action occurs amid AI being one of the most debated topics within the music and entertainment sectors, highlighted by the concerns raised after the removal of Copyright Register Shira Perlmutter.
Perlmutter over the weekend.

While the change addresses some critics’ concerns, it hasn’t seemed to appease all the critics. Ed Newton-Rex, the founder of the nonprofit music advocacy group Fairly Trained (who said last week he’d be removing his music from the platform) tweeted Wednesday that the update
“doesn’t go nearly far enough.”

As their new terms state, they won’t train general AI models that mimic your voice or style. However, they allow for more common types of AI training—models based on your work which may not exactly copy your style yet could still rival you in the marketplace,” Newton-Rex noted. “To truly tackle these worries, the necessary adjustment is straightforward. The clause ought to simply declare ‘We will not utilize Your Content for training generative AI models unless we have received your explicit permission.’

Read Seton’s
full letter here.