Seven years after its release, Stick Figure's "Choice Is Yours" shot up the charts. The band was thrilled - until they looked closer at what was driving it. A significant portion of the song's viral momentum came from unauthorized AI remixes circulating across streaming platforms and social media. The band never approved them. No royalties flowed back to the artists.
They're now in an ongoing fight to reclaim control of their own catalog.
How a Seven-Year-Old Song Gets Weaponized
The remixes weren't made by professional producers. AI audio tools that can take an existing song and reconstruct it in a different tempo, style, or arrangement in minutes have become cheap enough that anyone with a laptop can generate convincing-sounding versions. The output gets uploaded to Spotify, TikTok, and YouTube under slightly modified titles, where it competes directly with the original for the same search result slots and autoplay queues.
When a listener searches for "Stick Figure" and the algorithm surfaces an AI remix above the original, those streams go to the remix. The band gets credit for a viral moment they didn't fully benefit from. The actual royalties - already fractions of a cent per play on streaming - get split across versions the artist has no stake in.
The legal path forward is slow and expensive. Copyright law covers reproduction and distribution of original recordings, but AI remixes frequently get classified as transformative works, which makes platform-level enforcement difficult to force. DMCA takedown requests work for individual uploads, but new versions get re-uploaded faster than requests can be filed. Without a major label's legal team, independent artists have few practical options.
The Pattern Extends Everywhere
Stick Figure is dealing with a version of a problem that writers, podcasters, and video creators are running into in parallel. Writers see AI-paraphrased versions of their articles ranking above the originals in search. Podcasters find their voices cloned in fake episodes. The economics hit hardest in music because streaming royalties are so thin - any meaningful diversion of listens has real financial consequences.
Partial technical solutions exist. Audio watermarking embeds an inaudible identifying signal in a recording, but it can be stripped during AI processing. YouTube's Content ID system catches direct copies but often misses heavily modified AI versions. Neither addresses the volume of new uploads.
The platforms have limited incentive to act aggressively. More content means more engagement. The artists - especially independent ones without institutional backing - absorb the cost of that inaction.