Spotify has recently updated its policy regarding AI-generated audio to tackle the influx of AI-generated content on its platform. In a recent blog post, Spotify provided updates on this initiative, although it has not yet fully achieved its goal of helping users avoid low-quality content. This year, Spotify has removed 50 million tracks flagged as "low-quality and junk" from its recommendations. Over the past six months, Spotify has made multiple updates to its recommendation system to promote more tracks from "real human artists."
The company's editors also do not include tracks identified as low-quality or junk when curating playlists.
These updates were made after Spotify conducted a careful analysis of users' behaviors regarding "skipping, ignoring, or reflecting songs they do not want to hear." The company found a clear trend of users rejecting low-quality or junk tracks. Spotify stated, "Since we made these changes, we have seen a significant decrease in user concerns about such content appearing in recommendations." The company added that it will need to continuously update its systems to address spammers who are constantly changing their strategies to evade rules. Spotify indicated that it utilizes multiple signals to help detect low-quality and junk content, including metadata, user feedback, upload patterns, and third-party AI detection tools.
Spotify's detection technology for low-quality content faces challenges
However, these detection technologies have limitations, and as generative AI systems advance, it will become increasingly difficult for these tools to identify low-quality content. Spotify mentioned that it has removed over 75 million "junk tracks" from the platform due to "serious or repeated violations of our rules." This figure is consistent with what was provided when the AI policy changes were announced in September 2025. Although a spokesperson told Music Business Worldwide that Spotify is still removing "content that seriously or repeatedly violates our rules," there has been no update on the 75 million figure.
Spotify seems to be in a dilemma. On one hand, it claims to be removing low-quality and junk content from recommendations; on the other hand, it is adding generative AI features to the platform. For example, Spotify allows users to remix and cover songs using AI tools. Some artists, widely regarded as fake, such as Sienna Rose and The Velvet Sundown, have over 100,000 monthly active listeners on the platform. Another artist named Breaking Rust made it onto the Billboard Digital Song Sales chart last year, currently boasting around 700,000 monthly active listeners.
A report from July indicated that the top ten AI artists on Spotify collectively earned about $2.4 million, which undermines the human musicians Spotify claims to promote.
Spotify needs to provide users with options to filter AI-generated content
Spotify recognizes that a significant portion of its users do not want to hear generated low-quality content, but the company has only confirmed the removal of "low-quality and junk" tracks from recommendations, not necessarily from the entire platform. If its true goal is to "put human artistry first," then removing all such content should be an easy decision. Spotify has also not provided users with a straightforward way to block this type of content. This month, Deezer announced it would offer users a complete option to filter out AI-generated audio by the end of the year. Bandcamp essentially prohibits AI-generated audio, while Tidal does not pay royalties for tracks that are entirely AI-generated.
If Spotify could provide users with an option to choose not to listen to AI-generated audio, it would be quite beneficial.
However, earlier this year, a Spotify spokesperson told the BBC that its "priority is to address the harmful uses of AI, such as spam and impersonation, rather than trying to filter music based on how it was generated." Nevertheless, Spotify should at least offer users an option to completely block tracks identified as low-quality and junk, as well as a way to filter out those that disclose AI usage in their statements. Spotify stated that artists submit thousands of AI declarations daily to clarify the role of technology in their uploads.
Users can view these declarations in the now playing view on mobile devices, and Spotify is adding more ways for users to "quickly and easily see whether artists have disclosed AI usage in the song creation process." Currently, thousands of virtual artists have also received the AI Persona label, and these fictional tracks do not appear in editor-curated playlists or in users' recommendations by default (unless users actively follow them).
Spotify stated that its threshold of 1,000 plays before paying track royalties helps reduce this influx. "Tracks that show a high likelihood of being AI-generated have ten times the chance of being below this threshold, making it harder for bad actors to siphon royalties from real artists through mass-generated music." Additionally, the company noted that there are currently over 300,000 human artists certified on the platform, covering over 99% of the content that users actually search for.
Spotify is also expanding a tool that allows musicians to approve which songs appear on their profiles and are listed as artists, a feature that will be available to all artists by the end of the year.