A song that sounds like Drake and The Weeknd has drawn a large online audience, but neither artist made it. Called “Heart on My Sleeve,” the track was created with AI by an artist using the name Ghostwriter. The case brings a practical question into view: how should music platforms and rights holders handle songs that imitate artists without their involvement?
A convincing imitation can travel quickly
“Heart on My Sleeve” had amassed over 250,000 Spotify streams and 10 million views on TikTok. Those figures show how far an AI-generated imitation can circulate, even when the musicians whose voices it resembles did not take part.
Drake and The Weeknd had not publicly responded to the song at the time of the report. Drake had, however, reacted to another AI model of his voice, used to sing “Munch” by Ice Spice. In an Instagram story, he wrote, “This is the final straw AI.” The source noted that he may have been joking, while also placing the reaction in a wider pattern of artists objecting to deepfake music.
That pattern includes disputes over AI-generated videos featuring Jay-Z and Eminem. In 2020, Jay-Z’s agency Roc Nation submitted copyright strikes against YouTube uploads of AI-generated Jay-Z deepfakes, but YouTube reinstated the videos. The source also reported that UMG issued a copyright strike against YouTube videos of Eminem rapping about cats.
Copyright questions remain unsettled
Existing copyright law does not provide specific guidelines for generative AI, according to the article. It says transformative parody can be permissible, but whether a work qualifies as transformative is open to interpretation. The distinction matters because an imitation may raise questions about both creative expression and the unauthorized use of an artist’s identity or work.
The source also points to a lack of case law setting a clear precedent. Many disputes have settled before reaching a judge, leaving fewer court decisions to clarify how existing rules apply. For artists, labels and platforms, that uncertainty makes it difficult to know where a particular AI-generated song falls.
UMG has taken steps to limit AI music that imitates its recording artists. A Financial Times report said the company asked streaming services such as Spotify to prevent AI companies from using its music to train their models. A UMG representative described preventing unauthorized use and stopping platforms from ingesting rights-violating content as a responsibility to artists.
Training data is difficult to trace and remove
Questions about AI music connect to a broader problem: once creative work enters a training dataset, removing it can be difficult. The article describes this as a challenge for artists trying to regain control over how their work is used.
Technologists Mat Dryhurst and Holly Herndon founded Spawning AI to help address that problem. One of its projects, “Have I Been Trained,” lets users search for their artwork and check whether it has been incorporated into an AI training set without their consent. The tool illustrates one way to make training data more visible, though the source does not suggest that searching alone can remove work from a model.
The article gives illustrator Greg Rutkowski as an example of the stakes. His detailed high fantasy artwork for franchises like “Dungeons & Dragons” became one of Stable Diffusion’s most popular search terms when it launched in September. Rutkowski had not consented to his work being used to train the algorithm. The concern is that once an image-making system makes a recognizable style easy to reproduce, an artist may have limited ability to restore the control they had before.
The platform decision is still ahead
At the time of the report, Ghostwriter’s track remained on Spotify, and it was unclear whether it would stay there. Ghostwriter and Spotify had not immediately responded to TechCrunch’s requests for comment.
AI music generators may offer artists new creative possibilities, but the article’s examples show why the technology also raises questions about permission, compensation and platform responsibility. A song can attract listeners while its relationship to the artists it imitates remains disputed. Until legal standards and platform responses become clearer, cases like “Heart on My Sleeve” leave artists and services navigating uncertain boundaries.