Platforms Turn to Behavioral Signals to Combat Synthetic Music Fraud

Computer tools can now make many songs very quickly.

Some of these songs may be uploaded to music platforms without clearly saying how they were made.

When people play them, they may receive money from the same pool that pays human musicians.

Platforms therefore want to find suspicious uploads before payments are made.

Bharath Kandati says platforms should examine information around a song, such as its metadata and upload patterns.

They can also study how listeners and accounts interact with the track.

Listening to every song or examining the audio itself is difficult because streaming compression can hide important clues.

A computer system can detect unusual signals, but it cannot decide by itself whether using AI was acceptable.

The proposed solution combines computer detection, behavior analysis and human judgment, supported by clearer rules.