SoundCloud is facing fresh scrutiny over how its platform rules describe the role of artificial intelligence in uploaded audio. The issue centers on a terms of use update that appears to give the company permission to use user content in connection with AI and machine intelligence technologies.
What changed in SoundCloud’s terms
The latest version of SoundCloud’s terms includes language saying users explicitly agree that their content may be used to inform, train, develop or serve as input to artificial intelligence or machine intelligence technologies or services. The terms say this use is connected to providing SoundCloud’s services.
The provision was spotted by tech ethicist Ed Newton-Rex. According to the source article, the terms were last updated February 7, while Newton-Rex said he checked the Wayback Machine and that the language seemed to have been added to the terms on 12th Feb 2024.
For creators, the wording matters because SoundCloud is built around user-uploaded audio. A broad AI clause can raise immediate questions about whether tracks, vocals, samples, stems, demos or other uploaded material could become part of a system that improves AI tools.
The terms also include a carve out for content covered by separate agreements with third-party rightsholders, such as record labels. The source article notes that SoundCloud has licensing agreements with indie labels and major music publishers, including Universal Music and Warner Music Group.
Why the wording drew concern
The concern is not only about whether AI is used, but about control. TechCrunch reported that it was not able to find an explicit opt-out option in SoundCloud’s web settings menu. SoundCloud did not immediately respond to a request for comment before later providing a statement.
That absence of a visible opt-out is central to the broader creator debate. Many users argue that policies involving AI training should require people to actively opt in, rather than placing the burden on them to find a way out. Others argue that creators should be credited and paid when their work contributes to AI training datasets.
SoundCloud’s language also lands in a sensitive area because music is a rights-heavy medium. Uploaded audio can involve multiple layers of ownership and permission, from performers to publishers to labels. The source does not say how SoundCloud would apply the clause to each of those situations, which is why the carve out for separate agreements is important.
For independent creators, the practical question is straightforward: what happens to content that is not governed by a major licensing agreement or a separate third-party rightsholder deal? The source article does not provide a detailed answer, but the existence of the clause was enough to trigger criticism and demands for clarity.
How SoundCloud explains its AI use
After publication, SoundCloud provided a statement saying it has never used artist content to train AI models. The company also said it does not develop AI tools or allow third parties to scrape or use SoundCloud content from its platform for AI training purposes.
SoundCloud said it has implemented technical safeguards, including a “no AI” tag on its site, to prohibit unauthorized use. The company framed the February 2024 update as a clarification of how content may interact with AI technologies inside SoundCloud’s own platform.
According to SoundCloud, examples of those use cases include:
- personalized recommendations
- content organization
- fraud detection
- improvements to content identification with the help of AI technologies
The company also said any future application of AI at SoundCloud would be designed to support human artists by improving tools, capabilities, reach and opportunities on the platform. It cited possible uses such as better music recommendations, playlist generation, content organization and fraudulent activity detection.
SoundCloud specifically said tools like those from its partner Musiio are used for artist discovery and content organization, not to train generative AI models. That distinction is important because the term AI can describe many different systems, from recommendation software to generative models that produce new media.
The wider platform shift
SoundCloud is not the only large platform adjusting its policies around AI. The source article places the change alongside a series of recent moves by content hosting and social media services.
In October, Elon Musk’s X updated its privacy policy to let outside companies train AI on user posts. Last September, LinkedIn amended its terms to allow scraping of user data for training. In December, YouTube began letting third parties train AI on user clips.
SoundCloud has also been adding AI-powered tools to its own ecosystem. Last year, the company partnered with nearly a dozen vendors to bring tools for remixing, generating vocals and creating custom samples to its platform.
In a blog post last fall, SoundCloud said those partners would receive access to content ID solutions to help ensure rightsholders receive proper credit and compensation. The company also pledged to uphold ethical and transparent AI practices that respect creators’ rights.
Taken together, these details show the tension now facing creator platforms. AI tools can help organize content, recommend music, detect fraud and expand creative workflows. But when platform terms use broad language, creators want to know exactly what is being used, who can access it, whether consent is required and how compensation or credit will work.
What creators can take from the dispute
The SoundCloud situation is a reminder that AI policy is increasingly embedded in ordinary platform terms. A creator does not need to use an AI tool directly for their uploaded work to become part of a broader technology debate.
Based on the source article, the key facts are narrow but significant. SoundCloud’s terms contain AI-related permission language. A public critic identified the clause and questioned it. TechCrunch could not find a clear web opt-out. SoundCloud later said it has not used artist content to train AI models and described the provision as tied to platform functions rather than generative AI model training.
That still leaves a larger trust issue. Creators often need platform reach, but they also want predictable limits on how their work is handled. As more platforms revise terms for AI, the difference between internal platform improvement and AI training will need to be explained in plain language, not only in legal documents.