Spotify announced that so far this year it has prevented more than 50 million AI-generated or spam-like tracks from appearing in user recommendations. The company noted that this does not necessarily mean all 50 million recordings were deleted from the platform; rather, these tracks will no longer surface in the same way within recommendation feeds.
In addition, Spotify said it has removed at least 75 million spam-like tracks from its service for severe or repeated policy violations. That figure highlights how much of a challenge mass-produced automated content poses for streaming services.
Why AI-generated music is a problem for streaming
Artificial intelligence is not inherently a problem and can serve as a creative tool. The issue arises when actors generate large quantities of music with AI and upload it for profit, or manipulate systems with automated plays, repeated uploads and networks of manipulated playlists. In those cases success is determined less by musical quality or genuine listener interest and more by how effectively algorithms can be gamed — which can make it harder for human artists to reach audiences.
Changes to Spotify's recommendation signals
Spotify said it has more closely reviewed tracks that listeners skip, ignore or explicitly indicate they do not want to hear. Those tracks were consistently classified as spam or low-quality, mass-produced AI content, which prompted changes to the recommendation system.
In the coming weeks the company will introduce a new badge that can appear next to artist profiles, provided the artist is determined to be a real person. To make that determination Spotify will consider multiple signals, including listener engagement and feedback, upload patterns, metadata and third-party AI detection tools. The company is also monitoring tactics such as mass uploads, duplication, SEO manipulation, spam playlist networks and artificially shortened tracks. Spotify says it continuously updates its systems as spam operators change tactics.
Spotify also reported that user complaints about unwanted music appearing in recommendations have fallen since implementing these measures.
Earlier steps and other restrictions
In 2024 Spotify introduced a rule requiring tracks to receive at least 1,000 plays per year to be eligible for royalties, a change that made it harder for bad actors to profit from mass uploads. The company has also begun labeling artist personas generated by artificial intelligence, and says it has verified more than 300,000 human artists who account for over 99 percent of searches on the platform.
Remaining challenges
Spotify’s measures represent meaningful progress in combating spam and AI-driven manipulation, but the company acknowledges they are not final solutions. Abuse techniques and AI tools continue to evolve, requiring ongoing adaptation of detection algorithms and policies.
A central question for the coming period is whether streaming platforms can protect human creators from manipulation while still allowing creative uses of new AI technologies.
