SoundCloud says it isn’t using your music to train generative AI tools

The music-sharing level SoundCloud softly updated its presumption of usage successful February past year, adding connection that lets it bid AI models connected its users’ content, arsenic TechCrunch reported. And portion the institution says it hasn’t utilized user-created contented for exemplary training, it doesn’t regularisation retired the anticipation that it volition successful the future.

Marni Greenberg, SVP and caput of communications astatine SoundCloud, provided the pursuing successful a connection emailed to The Verge.

SoundCloud has ne'er utilized creator contented to bid AI models, nor bash we make AI tools oregon let 3rd parties to scrape oregon usage SoundCloud contented from our level for AI grooming purposes. In fact, we implemented method safeguards, including a “no AI” tag connected our tract to explicitly prohibit unauthorized use.

Greenberg went connected to accidental that SoundCloud’s presumption of work update “was intended to clarify however contented whitethorn interact with AI technologies wrong SoundCloud’s ain platform.” She said the institution uses AI for things similar personalized recommendations and fraud detection, and suggests its plans for aboriginal uses of AI connected its level autumn on akin lines.

When we asked astir letting users opt retired of having their euphony utilized for generative AI development, here’s what Greenberg had to say:

The TOS explicitly prohibits the usage of licensed content, specified arsenic euphony from large labels, for grooming immoderate AI models, including generative AI. For different types of contented uploaded to SoundCloud, the TOS allows for the anticipation of AI-related use. 

Importantly, nary specified usage has taken spot to date, and SoundCloud volition present robust interior permissioning controls to govern immoderate imaginable aboriginal use. Should we ever see utilizing idiosyncratic contented to bid generative AI models, we would present wide opt-out mechanisms successful advance—at a minimum—and stay committed to transparency with our creator community.

Hopefully SoundCloud volition spell to greater lengths to archer users astir those opt-out mechanisms than it appears to person done for past year’s AI-related presumption of usage update. Tech ethicist Ed Newton-Rex, who spotted the changes reported by TechCrunch, posted that they “can’t spot immoderate emails” alerting them that the presumption had been altered. I’ve contributed to SoundCloud, too, and besides didn’t find immoderate emails astir the changes erstwhile I checked. SoundCloud’s presumption say it volition supply “prominent notice” astir important alterations to its terms, but doesn’t warrant you’ll spot that successful an email.

Here’s SoundCloud’s archetypal afloat statement, arsenic provided to to The Verge by Greenberg:

SoundCloud has ever been and volition stay artist-first. Our absorption is connected empowering artists with control, clarity, and meaningful opportunities to grow. We judge AI, erstwhile developed responsibly, tin grow originative potential—especially erstwhile guided by principles of consent, attribution, and just compensation.

SoundCloud has ne'er utilized creator contented to bid AI models, nor bash we make AI tools oregon let 3rd parties to scrape oregon usage SoundCloud contented from our level for AI grooming purposes. In fact, we implemented method safeguards, including a “no AI” tag connected our tract to explicitly prohibit unauthorized use.

The February 2024 update to our Terms of Service was intended to clarify however contented whitethorn interact with AI technologies wrong SoundCloud’s ain platform. Use cases see personalized recommendations, contented organization, fraud detection, and improvements to contented recognition with the assistance of AI Technologies.

Any aboriginal exertion of AI astatine SoundCloud volition beryllium designed to enactment quality artists, enhancing the tools, capabilities, scope and opportunities disposable to them connected our platform. Examples see improving euphony recommendations, generating playlists, organizing content, and detecting fraudulent activity. These efforts are aligned with existing licensing agreements and ethical standards. Tools similar Musiio are strictly utilized to powerfulness creator find and contented organization, not to bid generative AI models.

We recognize the concerns raised and stay committed to unfastened dialogue. Artists volition proceed to person power implicit their work, and we’ll support our assemblage informed each measurement of the mode arsenic we research innovation and use AI technologies responsibly, particularly arsenic ineligible and commercialized frameworks proceed to evolve.