Nearly four in ten tracks released worldwide in July involved AI in some way, according to an analysis by music promotion service SubmitHub reported by EDMTunes. It ran more than a million July releases through its own detector, SH Labs, and classed 23.2% as fully AI-generated and another 15.3% as AI audio that people had later edited or processed.
SubmitHub raised its own caveat: when it asked artists about tracks its tool had flagged, 31% said they had used no AI at all, leaving open whether that reflects confusion about what counts or errors in detection. Speaking to Mixmag, founder Jason Grishkoff said disclosure matters most, so listeners can decide for themselves whether to engage with AI music.
The study lands as Bandcamp and Traxsource adopt detection tools, as rights bodies watch the copyright questions around training data, and as major and independent labels propose chart rules under which only substantially human-made tracks, using authorised AI services, could qualify.
Sources · 1