Tidal to Tag and Stop Paying Royalties on AI-Generated Music

Tidal implements strict AI music policy

Tidal announced it will begin tagging music identified as 100% AI-generated and will stop paying royalties on those tracks starting mid-July. The streaming service also plans to require labels and distributors to disclose AI-generated content when uploading music to the platform.

The policy represents one of the most stringent approaches among major music services as the industry grapples with an unprecedented surge in AI-generated content. According to data from Deezer, the platform was receiving an average of 170,000 tracks per day as of late March, with 75,000 (44%) detected as AI-generated. Deezer also identifies as much as 85% of AI track streams as fraudulent, generated by bot farms designed to siphon off royalty payments.

Why it matters

Music services face a fundamental challenge: while AI detection technology can identify purely AI-generated tracks with reasonable accuracy, it cannot distinguish between fraudulent content and legitimate artistic projects using AI tools. This limitation is forcing platforms to make broad policy decisions that affect both bad actors and human artists experimenting with AI as a creative medium. The policies these services adopt will shape how AI music develops commercially and whether artists can build sustainable audiences around AI-assisted work.

Detection technology becomes infrastructure

AI detection technology is rapidly becoming part of the essential infrastructure layer of digital music, alongside automated content recognition and industry data standards. Larger services like Spotify and Apple Music have developed detection capabilities in-house, primarily for spam and fraud prevention.

A growing vendor ecosystem has emerged to serve the market. Deezer built its own detection system and now licenses it to third parties. Ircam Amplify, the commercial arm of Paris's IRCAM research institute, offers detection products, while startups including HumanStandard, SoundSafe.ai, and Detect.Music have entered the space. Established vendors like Pex and Beatdapp have added AI detection to their existing technology stacks.

These systems report confidence levels that a track was purely AI-generated and are calibrated to minimize false negatives—incorrectly identifying human-created music as AI—over false positives. However, they cannot detect AI assistance in production tasks or identify tracks where only some instruments were AI-generated.

Divergent platform policies

Music services have adopted widely varying approaches to AI content. Bandcamp relies on user reports to enforce a no-AI policy, investigating and potentially removing tracks when users complain. TuneCore restricts submissions to AI music created with platforms trained on licensed datasets, while CD Baby bans all fully AI-generated music and terminates repeat offenders.

Most services accept AI music submissions but differ on disclosure requirements, user-facing tags, algorithmic promotion, and royalty payments. Spotify and Apple Music have developed voluntary AI disclosure tag sets for uploaders, though these standards differ from each other and from the DDEX industry standard.

Deezer became the first service to display "AI-generated content" tags in its user interface in June 2025. Both Deezer and audiophile-focused Qobuz have opted not to promote detected AI tracks through algorithmic playlists or search results. Qobuz and Tidal have taken the additional step of refusing to pay royalties on detected AI music, citing the lack of copyright protection for purely AI-generated works under U.S. law.

Smaller services have generally adopted more restrictive policies, positioning themselves as artist-friendly alternatives while managing limited resources for handling the growing volume of AI content and fraud.

Arms race ahead

The technology and policies will continue evolving as the industry gains experience with AI music, fraudsters develop evasion techniques, and AI tools grow more sophisticated. Services claiming to modify AI tracks to evade detection have already appeared, though AI music platforms themselves have not attempted to make their outputs detection-proof.

The industry faces a similar trajectory to earlier technologies like digital rights management and content recognition, which went through periods of market consolidation and ongoing arms races with bad actors developing workarounds.

These details were first reported by Bill Rosenblatt in Forbes.