How streaming fraud quietly steals from real artists

The 2024 to 2026 inflection is AI. Tools now generate tracks faster than humans ever could, and fraudsters flood the catalogue with AI songs streamed thinly by bots to skim royalties under the radar. Industry bodies put the drain at around two billion a year, roughly three to ten per cent of the entire streaming royalty pool.

How the bots do it

Botnets use residential proxies and automation to look like genuine, geographically scattered listeners, and they play each track just past the thirty-second mark, the point most services count as a paid stream. They spread the streams thinly across hundreds of thousands of tracks so no single one shows a suspicious spike. The flood is real: one major service reported that around a quarter of its daily uploads were fully AI-generated, tens of thousands of tracks a day, though those AI tracks still made up only a sliver of actual listening.

What the platforms are doing, and the human cost

The services have started to bite back, with minimum-stream thresholds before a track earns anything, per-track penalties charged to the distributors of fraudulent uploads, and large purges of spam tracks; one reported demonetising around two billion fraudulent streams in a year. The most prominent case so far, a producer who used AI tracks and bots to fake billions of streams, ended in a guilty plea and a multi-million forfeiture. The limit and the lesson are the same: the dollar figures move and should be dated, but the shape does not. The fraudster's audience is bots, the loss lands on real artists, and the only thing fakery here builds is a bill someone else pays.