Spotify says it excluded more than 50 million slop and spam tracks from recommendations during 2026, while leaving those recordings available in its catalog. The October 20 update describes changes intended to reduce unwanted material and large-scale abuse, alongside AI disclosures and artist protections. The figure does not mean that every affected song was entirely AI-generated or that all were deleted.
Spotify has stopped recommending more than 50 million tracks identified as slop or spam during 2026. The figure, disclosed on October 20, describes a change in what its systems place before listeners, rather than a mass deletion of every affected recording. The tracks remain available to people who deliberately look for them, but lose automatic exposure through the service’s recommendation mechanisms.
That distinction separates access from promotion. A recording can exist in the catalog without being selected for personalized listening or editorial programming. For artists trying to reach unfamiliar listeners, those surfaces can be significant routes to an audience. Removing a track from them changes its chances of being discovered, even if a direct search can still bring someone to the same recording.
Spotify began by examining tracks people repeatedly skipped, ignored or explicitly said they did not want to hear. It found a consistent overlap with material its systems had identified as slop or spam. Those listening signals helped guide changes during the previous six months, linking the intervention to how people responded to recommendations rather than relying solely on an assessment of the audio.
Spotify also said its editors avoid programming material identified in those categories. The policy therefore applies to human selections as well as automated discovery. It reported a marked fall in listener concerns after the changes, but did not give a percentage or publish an independent before-and-after measurement that would quantify the improvement for an outside observer.
The 50-million figure should not be read as a count of songs proven to have been created entirely by AI. It groups slop and spam together without a published breakdown separating every production method. The term AI slop describes mass-produced, low-quality material, but unwanted distribution practices can involve more than one kind of recording and do not become identical simply because they share a filter.
No single signal identifies the problem reliably. The service combines listening feedback with upload patterns, release metadata and information from third-party AI detection tools. It acknowledges limits in detection technology and expects those difficulties to grow as generative systems improve. Combining indicators is intended to support a more confident judgment than treating one tool’s result as a complete answer.
Practices under examination include mass uploads, duplicates, misleading search-oriented metadata, spammy playlist networks and artificially short tracks. These can seek attention or streams through volume and manipulation rather than through ordinary listener interest. Spotify says it continues looking for new tactics as uploaders change their methods, making the filtering process an ongoing task rather than a one-time catalog cleanup.
A separate figure concerns more than 75 million spam recordings removed from the platform altogether. That number was already disclosed in 2025 for serious or repeated rule violations. It is not an additional 75 million newly deleted tracks attributed specifically to 2026. Recommendation exclusion and full removal are different actions, with different consequences for whether someone can still access the recording.
AI Persona badges add another layer of information. They identify profiles whose public artist identity does not represent a real person. Thousands of such profiles already carry the designation. Their music is generally excluded from automated and editorial recommendations, with exceptions such as a listener deliberately following that identity. The label describes the profile and its treatment without amounting to a universal ban on listening.
The company also allows creators to disclose which parts of a recording used AI. These credits can identify particular roles or instruments, giving listeners more detail than a simple division between wholly human and wholly generated music. Tens of thousands are submitted daily. Spotify presents disclosure as a transparency tool, separate from deciding that a track constitutes spam or violates its rules.
For artists, a Spotify pilot program offers more control over releases presented under their names. Thousands have joined a process intended to help them review incoming material and stop unrelated or misleading uploads reaching an official profile. That addresses impersonation through distribution, including situations where somebody submits altered recordings under a recognizable artist’s name and creates confusion about whether a release is genuine.
Official verification covers more than 300,000 artist profiles, the company said, representing over 99 percent of artists users search for. More than seven in ten verified artists are independent. The badge supplies another signal of identity, while the profile-review pilot addresses a different problem: deciding whether a particular release belongs with the artist to whom it has been delivered.
The economic rules remain distinct as well. Tracks must reach 1,000 streams over the preceding twelve months, alongside other conditions, to enter the recorded-music royalty calculation. Introduced in 2024, that threshold is not a blanket AI exclusion. The October update says recordings with strong signs of AI generation are ten times more likely to fall below it. For listeners, the immediate change is simpler: the identified material can still be searched, but no longer occupies the same recommendation spaces.
