X-rated, AI-generated country songs are taking over the internet

IT SOUNDS LIKE a typical country tune, for a few seconds at least. “Country Girls Make Do” opens with a toe-tapping beat and twangy strains of fiddles and guitar. You expect a song about whisky or worship—but what follows is downright unchurchly. When the singer announces that she is “horny as a bitch in the country heat”, and handling ears of corn in ways that no farmer intended, you realise you are listening to something much filthier than Red Dirt music.

The song is deliberately outlandish, which may be why “Country Girls Make Do” has gone viral in recent weeks, racking up millions of streams and likes across various platforms. In one TikTok trend, people trick their partners into dancing to it. In another, youngsters play the song to their relatives, claiming that it is the latest chart-topping hit, and record their reactions for laughs. “Is this for real?” one horrified grandmother asks.

Yes and no. “Country Girls Make Do” is a parody. What makes it unusual is that it has been created by artificial intelligence (in this case an “artist” called Beats By AI). On apps such as Suno and Udio users can input a few prompts—whether PG or X-rated is up to them—and the programme will generate lyrics, vocals, instrumentation and even artwork. Melodies can be refined and pitches perfected, with no real music knowledge required.

Beats By AI is one of many such acts. Aventhis is a brooding country outfit which uses AI for instrumentation and vocals; it has more than 650,000 monthly listeners on Spotify. The Velvet Sundown, whose 1970s-inspired rock tunes attract almost 300,000 monthly listeners, is fully AI-generated. AI can produce music swifter than Taylor Swift ever could. Both Aventhis and The Velvet Sundown have released three albums this year. Beats By AI releases a tawdry track every day.

AI songs are catching on because they offer familiar sounds. In the case of Beats By AI, part of the music’s appeal is its send-up of country tropes. For the most part, the genre remains stubbornly conservative—record labels in Nashville are known for dictating a certain style and sound—yet its growing Gen Z fan base is anything but.

“I’ve been waiting for this to happen,” says Daniel Gervais, a professor of intellectual-property law at Vanderbilt University. He sees these songs as the first stage of a “revolution” whereby consumers generate the content they want instead of settling for what record labels package and sell. Anyone with the time and the inclination can make bangers to match their mood or to gleefully defy the industry rules.

Musicians, however, are not amused by AI-generated songs, no matter how funny their lyrics. Hitmakers including Sir Elton John, Dua Lipa and Ed Sheeran have expressed their fears of AI companies drawing on their work to train models without proper remuneration. And streaming services are not all that happy about it either. On Deezer and Spotify fraudsters are uploading AI-generated music, using bots to “listen” to it and raking in royalties. Deezer has removed such tracks from official playlists and algorithmic recommendations. In September Spotify purged its platform of 75m AI-generated songs: “Bad actors”, the firm explained, were pushing “slop into the ecosystem”.

Plenty of music lovers, though, do not mind the muck. Luminate, a data firm, has found that one in three American listeners is “somewhat” or “very” comfortable with AI being used to write instrumentals. (Lots of professional musicians use AI tools in some way already.) Almost 30% are happy for machines to write lyrics and more than 25% do not mind an AI singer performing an AI-composed song. “I take back everything I’ve [said] about AI art,” one person wrote under the “Country Girls Make Do” video on TikTok. “Let it cook.” ■