If you had TikTok in Autumn 2025, I’m willing to bet my student loan that you’ve heard the song ‘I Run’ by Haven. It was everywhere, even getting to number 25 in the Spotify global chart, having gained popularity largely because the singer sounded remarkably like British singer Jorja Smith.
Then people realised the vocals were AI. The song suddenly disappeared overnight, was quickly condemned online and purged from the charts.
Partly or fully AI-generated songs like ‘I Run’ are becoming increasingly common, thanks to readily available audio tools like Suno or Mureka, and they can be annoyingly hard to spot. And once you know AI can be used in music, it’s hard not to suspect that it’s everywhere.
So, how do you spot an AI-generated song? Spotify, for example, has added a handy ‘AI persona’ badge to artists who self-disclose as or are suspected of being AI, but there’s so much AI-generated music that much of it goes unchallenged. One way of sorting the real from the fake is to look for clues like generic song structures that indicate something may not be entirely human produced. Polished production is also a telltale sign of AI music, missing the imperfect human aspects – voice cracks, wrong notes, slip ups – that characterise a lot of the songs we listen to.
This is not to say that there are no hypothetical benefits of using AI in music. It can, for example, be used to streamline production so that artists can focus on more creative aspects. But it’s also important to point out how AI’s rise has harmed the music industry despite these benefits.
Al-generated music ‘slop’ has been a source of entertainment online, and increasing AI usage has brought new challenges to artists’ rights. One of the biggest issues is that generative AI uses existing music to train itself. The technology has outpaced regulation in this sense, with the Musician’s Union still fighting for new securities in areas from data protection to revenue sharing. Artists in the UK currently lack concrete ways to protect themselves in the industry as these tools continue their meteoric rise.
But even aside from these regulatory issues (and there are many), is AI-generated music even enjoyable? How would an AI perform live at a concert? Is music as fun to listen to when you know that very little human work has gone into making it? The backlash against songs when they’re revealed to use AI like ‘I Run’ suggests that it isn’t for most people, at least not yet.
The rise of AI in music has wide-reaching implications for both artists and listeners. For me, knowing something has used AI immediately makes it much less entertaining. But maybe I’m stuck in old ways, and this really is the future of music. If so, more regulation is needed, quickly.
