Spotify said Oct. 20 that it has removed more than 75 million spammy tracks from its platform and taken more than 50 million tracks out of recommendations. The same day it signed Believe to a licensing deal for a planned paid tool that lets fans make AI-powered remixes and covers.
Spotify’s Spam Removal Numbers
Spotify’s one-year review of its AI policies reports two removal counts. It took more than 50 million “slop and spam” tracks out of recommendations this year. It removed more than 75 million spammy tracks from the platform entirely for serious or repeated violations of its rules. Added together, the two figures come to more than 125 million, though Spotify has not said whether any track falls in both counts.
Spotify put the size of its catalog at over 100 million tracks in its Believe announcement. Measured against that figure, the 75 million removals equal up to 75% of the catalog’s current size. The company has not published the catalog total before the removals.
Spotify also said tracks with indicators of a high probability of AI generation are 10 times more likely to fall below its 1,000-stream royalty threshold. That threshold has been in place since 2024, and Spotify says it keeps royalty money from flowing to tracks built only to collect it.
How Spotify Detects Slop and Spam
Spotify said it reviewed skipped and ignored tracks and found most carried indicators of slop or spam. Over six months it changed its recommendation system to remove such tracks and surface more human-made music. Its editors do not program content identified as slop or spam, and the company said listener complaints about the material have fallen markedly.
The company combines several signals: listener engagement and feedback, upload patterns, metadata and third-party AI detection tools. It tracks tactics such as mass uploads, duplicates, search-engine manipulation, spammy playlist networks and artificially short tracks. Spotify acknowledged that AI detection has limits. The post gives no error rate for the detection, no count of tracks removed by mistake and no appeals figure.
Labels and Protections for Artists
Spotify set out its plans on Sept. 25, 2025, and described four tools it has launched since. AI Credits, launched in April, lets artists disclose which roles and instruments used AI. Listeners see the credits in Song Credits in the Now Playing view on mobile, and Spotify said tens of thousands of credits come in daily.
Verified by Spotify, launched April 30, is a badge confirming that a real, identifiable person made the music. Spotify said more than 300,000 verified artists represent over 99% of what listeners search for, and more than 70% of them are independent.
AI Persona badges, announced Aug. 11, label artist identities that Spotify believes are AIgenerated. Spotify excludes AI Personas from editorial and algorithmic recommendations by default unless a listener follows them. Thousands have been labeled, and Spotify said it is the only major streaming service that labels them.
Artist Profile Protection lets artists review releases that list them before those releases appear on their profile. Thousands of artists joined the beta, and Spotify said it will reach all artists by the end of 2026. The post’s guiding principle is that “human artistry comes first.”
The Believe Remix Tool and How Royalties Work
Artists and labels from Believe and TuneCore can benefit from the licensing agreement for Spotify’s planned tool. Spotify said the tool will be a paid add-on for Premium subscribers. Fans will make AI-powered remixes and covers of existing songs, not entirely new works. Every remix links back to the original track and artist. Streams of remixes generate royalties for participating artists and for the rights holders behind the original works, on top of what they already earn on Spotify.
Participation is voluntary and uses two steps. Artists first opt in to being considered for AIrelated programs. Opted-in artists then decide whether to join the Spotify partnership. Believe says its responsible AI principles, set in 2023, rest on consent, control, compensation and transparency.
Denis Ladegaillerie, Believe’s founder and chief executive, said in the announcement that AI “must create new opportunities for artists” while protecting their rights. Charlie Hellman, Spotify’s senior vice president and global head of music, said in the same announcement that fan creativity done right can strengthen the bond between fans and artists. Believe said it is building on its generative AI music framework and distribution safeguards and will release more details later. Spotify did not give a launch date or a price for the add-on. It did not say how it will split remix royalties between an artist and the owners of the original song.
Earlier Licensing Deals and Company Size
The Believe agreement follows deals the announcement lists with Universal Music Group on May 21, Merlin on Aug. 4 and Kobalt on Aug. 13. Believe operates in more than 50 countries and has more than 2,000 employees. Spotify reports 777 million users, including 300 million subscribers, in 184 markets. Its catalog includes more than 7 million podcast titles and 700,000 audiobooks along with the music.
Spotify published the review and the Believe announcement the same day. The Believe post says Spotify is strengthening protections against spam, impersonation and deception, with the aim of keeping AI-generated spam out of recommendations and giving artists more control over their profiles. It also says listeners can see when artists have disclosed AI use in a track.
The deals show how Spotify is pairing two kinds of policy. One removes tracks it considers spam. The other licenses AI-assisted derivative works from artists who agree. The 300 million subscribers are the audience for the paid add-on, and Spotify has not said how many will pay for it.
