AI music floods streaming platforms, prompting artists to clarify their human status

Listeners on streaming services are increasingly encountering music created by artificial intelligence rather than humans. This shift forces consumers to question the origin of the tracks filling their playlists. Some artists have faced public scrutiny and legal pressure over their use of AI tools, while others have built substantial audiences without clearly identifying their work as synthetic.

The issue came into sharp focus with the song “I Run” by the artist HAVEN. The track gained popularity on TikTok nearly a year ago, with hundreds of thousands of videos using the audio. Users eventually noticed the vocals sounded unnatural and the audio quality was fuzzy. HAVEN later admitted to generating the song using Suno, an AI software that creates tracks from text prompts. The vocals were reportedly trained on the voice of English singer Jorja Smith without her permission. Smith threatened legal action, which led to the track being removed from Spotify, where it had been one of the highest-charting songs. The song was later rerecorded with vocals from human artist Kaitlin Aragon and now has over 260 million streams on the platform.

Other AI-generated acts are maintaining large followings without such dramatic interventions. Two Spotify profiles, The Velvet Sundown and Kayla Kross, have each accrued over 100,000 listeners and millions of total listens. The Velvet Sundown describes its project in its biography as an exploration of human-guided synthetic music. Kayla Kross’s profile, however, is written as if she were a traditional human artist. Her bio uses descriptive language such as “choosing tenderness as an act of courage” without explicitly stating that the music is machine-made.

The financial implications of these streams are real for the entities behind the profiles. With millions of plays, these accounts likely generate revenue for the people or companies managing them. The lack of consistent disclosure makes it difficult for listeners to know if they are supporting a human creator or a digital catalog. Some observers argue that presenting AI-generated work as human-made crosses the line from artistic persona into deception, especially when the primary goal is to profit from prompted content.

The technology behind these tracks has improved rapidly. A comparison between the 2023 track “Heart on My Sleeve,” which cloned the voices of Drake and The Weeknd, and recent releases like Kayla Kross’s “Let You Feel This Love” shows notable gains in instrumental quality. While AI vocals still possess a cloudy quality, they are sharper than they were three years ago. The industry expects these improvements to continue as generative models advance in other media fields such as image and video generation.

As AI music becomes more prevalent, distinguishing it from human-made work is becoming more challenging. One common indicator is the way AI models handle sibilant sounds, such as “sss” noises in lyrics, which often sound less human-like. However, this tell may become less reliable as models improve. Listeners may need to become more diligent in researching the origins of the music they consume to decide if they prefer to support human-created art.