Music streaming giant Spotify said Tuesday it had removed more than 50 million tracks flagged as AI-generated low-quality content or spam from its recommendations this year, stepping up efforts to protect listeners and support human musicians.
In a statement marking one year since introducing stronger protections against AI-related spam and impersonation, Spotify said it had also removed more than 75 million spam tracks from its platform for serious or repeated violations of its rules.
Unlike tracks deleted from the platform, the 50 million excluded from recommendations were not necessarily removed from Spotify’s catalog.
“AI-generated slop and spam don’t belong in your recommendations, and we’ve made changes to keep them out and protect the listening experience,” the company said.
Spotify said it had updated its recommendation system over the past six months after identifying patterns among tracks that listeners frequently skipped, ignored or reported as unwanted.
To detect suspicious content, the company combines listener feedback, upload patterns, track metadata and third-party AI detection tools, while acknowledging the limitations of AI detection technology.
The company also highlighted measures introduced over the past year, including AI Credits that disclose the use of artificial intelligence in music production and badges identifying AI-generated artist identities.
More than 300,000 artists have received Spotify verification badges, while thousands of AI-generated artist identities have been labeled and excluded from recommendations by default.
Spotify said tracks considered highly likely to be AI-generated were 10 times more likely to fall below its annual 1,000-stream threshold for royalty payments, making large-scale attempts to exploit its payment system less profitable.
The company said it would continue updating its detection systems as generative AI technology evolves and spammers develop new tactics.
