When ‘Heart On My Sleeve’ was quietly released on streaming services on 4 April 2023, one wonders whether its creator Ghostwriter977 could have anticipated that it would precipitate a watershed moment for the music industry. 

Discussions around the implications of the nascent AI boom had been percolating in the background for some time before the track was released. Much head-scratching has gone on at music conferences where AI has become a dominant topic on panels exploring everything from creator rights to music production, and predictions have been made in countless meetings behind closed doors with growing concern.

In the days that followed the release of ‘Heart On My Sleeve’, Universal Music Group (UMG) pulled the track from YouTube, Spotify, Apple Music and other streaming services, condemning it for infringing copyright, and stating that platforms have a ‘fundamental legal and ethical responsibility to prevent the use of their services in ways that harm artists’. 

The track’s targeted removal and UMG’s stern statement suggested its level of success represented a real concern. Prior to the initial removal effort, ‘Heart On My Sleeve’ racked up 600,000 streams on Spotify, 275,000 views on YouTube and 15 million views on TikTok, with one unofficial Twitter reupload racking up 6.9 million views. In the days after 16 April, when the track blew up on TikTok, the trend was almost impossible to miss. 

However, while ‘official’ versions of the (unofficial) track may have been taken down, we at Song Sleuth became curious as to the ripple effect created by the track within the user-generated content ecosystem. So we took it upon ourselves to go on a deep-dive, and this is what we found.

Key takeaways:

• An average of 35 UGC videos containing the ‘Heart On My Sleeve’ audio or direct derivatives were uploaded to YouTube per day for at least the first 2 weeks, from 17 April until 1 May.
• UGC included reactions, instrumentals, sped-up versions, remasters, instrumental covers, and even human rappers ‘covering’ the track. 
• Rightsholders made an aggressive effort to have content taken down when the trend was new, but maintenance represented an ongoing challenge. Over half of the infringing videos have remained on YouTube at all times:
• 54% (20 of 37) of Drake AI videos uploaded on 17 April were still available as of 13 May, while fully 91% (30 of 33) of videos uploaded on 30 April, just 2 weeks later, were available to view.
• Overall, lifetime views for the YouTube videos uploaded on 17 April alone totaled 389k (as counted on 13 May, among videos still online). Views were largely enjoyed by already-popular channels, and views did drop off steeply among uploaders later to the trend. 

What we can clearly see from our analysis is that UGC around AI-generated tracks is a new trend that creators are already capitalising on. And in the case of Fake Drake, even despite rightsholders’ best efforts to take down content on YouTube, the majority of the content was – and continues to be – missed.

It is undeniable that AI represents a new challenge for the music industry on many levels from legal to artistic as AI covers continue to proliferate. But when it comes to AI-assisted creation and the world of UGC, Song Sleuth is able to help monitor and gather data at scale through our unique proprietary technology. How the industry will navigate generative AI from here remains to be seen, but we will certainly be here to help rightsholders claim and monetise all forms of their content as we move through the new landscape collectively.