AI learned to sing in Spanish. Latin artists need to get paid.
An opinion piece argues that generative AI trained on massive music datasets threatens Latin artists by cloning voices and competing with human-made recordings. The author highlights unauthorized use of catalogs in model training, rising AI-generated uploads on streaming platforms, legal responses and industry deals that show consent-and-pay approaches are possible.
Why It Matters
Latin music is the world’s most streamed genre family and recently surpassed $1 billion in U.S. revenue, so widespread AI cloning or uncompensated use of Latin catalogs could disproportionately cut creators’ income. Policy, record-label action and contract language made now will shape whether musicians retain control and revenue from their voices.
Key Facts
- Incident prompting concern: A reggaetón track on TikTok used a convincing AI clone of Bad Bunny's voice alongside false vocals attributed to Daddy Yankee and Justin Bieber, prompting Bad Bunny to warn fans.
- Dataset investigation: The Atlantic found AI training datasets containing roughly 21.2 million tracks, including material from Bad Bunny's catalog, allegedly ingested without authorization.
- Lawsuits: Universal Music Group and Sony Music have sued the AI platform Suno after identifying tens of thousands of their recordings in its training data.
- AI-generated uploads: Deezer reported that nearly half of its daily music uploads are now AI-generated.
- Latin music revenue milestone: In 2025, Latin music revenue in the U.S. exceeded $1 billion for the first time, marking a tenth consecutive year of growth.
The rise of generative audio tools has created a new battleground for Latin music creators, who provided much of the raw material used to train voice- and style-cloning models. High-profile examples — such as a deepfake reggaetón track that mimicked Bad Bunny’s voice — have focused attention on how AI systems were trained and whether artists were asked or paid before their recordings were included.
Journalistic and industry inquiries have revealed the scale of the data feeding some models: an investigation by The Atlantic found roughly 21.2 million tracks inside training datasets, reportedly including recordings by major Latin artists. Major labels have responded legally; Universal Music Group and Sony Music filed suits against the AI company Suno after discovering tens of thousands of their tracks in its datasets. At the same time, streaming platforms are seeing synthetic content proliferate — Deezer says about half of its daily uploads are now AI-generated — creating fresh competition for playlists, listener attention and royalty pools.
The stakes for Latin music are particularly high because the genre family is the most streamed worldwide and recently crossed $1 billion in U.S. revenue in 2025. Industry research cited in the piece warns that generative AI could shave a substantial portion of creators’ earnings; a study commissioned by the International Confederation of Societies of Authors and Composers projects creators might lose up to nearly one-quarter of income to AI by 2028, a cumulative impact of roughly €10 billion.
Responses are emerging on multiple fronts. In the U.S., lawmakers reintroduced the NO FAKES Act in May to grant people federal control over AI replicas of their voice and likeness, including takedown procedures. Some AI firms have chosen licensed paths: Suno announced new models built under agreements with Warner Music Group and BMG that include payment for participating artists, illustrating an alternative to scraping catalogs without consent. The author argues that practical protections will also require contract provisions explicitly covering AI training, persistent metadata, and collective organization by artists across borders, since models trained in one country can affect performers worldwide.
Keep Reading
What is the AI singularity? Some experts say it’s closer than ever
TechCrunch Mobility: How do we know when an AV is safe enough?
Vocci’s ring adds a new form factor to meeting note-taking
