rotative
theverge ai · June 20, 2026

The Atlantic found 21M songs training AI — is your music in them?

the atlantic launched a searchable database of over 21 million songs found in ai training datasets, revealing that major artists like lady gaga and radiohead are included without explicit licensing.

read in 简体中文 · 繁體中文 · 日本語 · 한국어

the atlantic released a searchable database exposing four datasets containing over 21 million songs used to train ai models, ranging from pop stars like lady gaga to experimental acts like aphex twin. alex reisner found that while some datasets like the free music archive allow personal streaming, commercial ai training requires licensing that likely hasn't been secured for most tracks. google and stability ai have confirmed using these datasets in research, despite the fact that the scraping methods often bypass youtube and spotify monetization mechanisms.

read at source ↗

← rotative