Inside Spotify's AI Music Crackdown: 75 Million Tracks Removed, New Policies, and the Future of Royalties

New policies and enforcement pillars

The September 2025 policy update established three enforcement pillars :

• Improved impersonation policy — Clarifies how Spotify handles claims about AI voice clones and other unauthorized use of an artist's name or likeness . The policy targets "deepfakes" uploaded to artists' official profiles .
• Upgraded spam detection system — A new automated music spam filter that, by mid-2026, was auto-flagging approximately 39% of all uploads . Tracks are flagged before or at upload; the system also catches artificial streaming manipulation .
• AI disclosure requirements — Creators must now disclose AI use through industry-standard metadata. The system uses the DDEX 5.0 standard, which ships with mandatory AI-disclosure fields . Disclosure is voluntary and Spotify has stated it will not downrank or punish "responsible AI use" .

Transparency measures

The most concrete transparency step was the AI credits beta, which launched on April 16, 2026 :

• Artists and labels can declare, via DDEX metadata, whether AI was used in vocals, instrumentation, or post-production.
• DistroKid was the first distribution partner to support the system .
• A small AI tag appears inside a track's credits view on mobile when disclosure is made .
• Listeners can see whether and how AI was involved in a track they are streaming .
• Apple Music followed with a similar "Transparency Tags" delivery requirement on March 4, 2026 .

Spotify also introduced a verified artist badge — AI personas and fully AI-generated artists are not eligible for the badge, which signals that a profile represents a real human artist .

Royalty system protection

Spotify pays from a shared royalty pool, so every fraudulent stream from AI-generated spam reduces payouts to legitimate artists . The platform's protections include:

• Clawing back fraudulent payouts — Spotify says it has recovered money from artificial streaming operations linked to AI-generated content .
• A 1,000-stream minimum monetization threshold per track, which blocks most low-engagement AI spam from ever reaching the royalty pool .
• The spam filter catches mass uploads and stream-manipulation schemes before they can inflate royalty payouts .

Listener engagement with AI-generated content

Independent data shows AI-generated music has very low listener engagement:

• An academic study published in June 2026 using Spotify's metadata (covering hundreds of millions of tracks and a recommendation graph of 33 million tracks) found that 93% of AI-generated music receives few or zero listener plays and is rarely recommended by Spotify's algorithms .
• Spotify itself has stated that fully AI-generated tracks see "a minimal share" of total streams and do not meaningfully affect streams or revenue for human artists .
• A Luminate report from May 2026 showed listeners are growing more dissatisfied with AI-generated content .
• On Apple Music, AI-generated content accounts for less than 0.5% of total listening time ; comparable data for Spotify has not been publicly broken out with the same precision, but the trend is consistent across platforms.
• By contrast, supply is exploding: by April 2026, AI-generated tracks made up 44% of all new daily uploads to Deezer (roughly 75,000 tracks per day), and the situation is similar on Spotify . Despite this flood, actual consumption remains negligible relative to human-made music.