Photo Credit: Markus Winkler
One guilty plea later, the federal wire fraud case against streaming scammer Michael Smith is a wrap. But have DSPs remedied the shortcomings that allowed the multimillion-dollar scheme to unfold?
Though not quite new, this and related questions are now taking center stage. Ahead of a scheduled October trial, Smith yesterday entered a guilty plea on one count of conspiracy to commit wire fraud and is facing a maximum of five years behind bars.
Additionally, the 54-year-old, having reaped millions in ill-gotten royalties by directing billions of bot plays to his AI uploads, agreed to pay nearly $8.1 million. With that, the presiding judge is set to sentence the streaming fraudster on July 29th.
However, as DMN Pro subscribers are aware – we were out first here, and some in the industry forgot to cite Digital Music News when repackaging exclusively reported details as their own – the episode doesn’t begin and end with Smith.
On the streaming side, the fraudulent royalties had to come from somewhere. And while the responsible platforms weren’t (and aren’t) eager to disclose their Smith payouts, those that promptly spotted the scam didn’t hesitate to let us know.
Spotify, Pandora, Deezer, and (probably) SoundCloud told us they’d quickly flagged the fake streams and pulled the tracks, thereby making inconsequential payments to Smith. On the other hand, Apple Music, Boomplay, Tidal, YouTube, and Amazon Music opted for the “radio silence” strategy when contacted for comment.
Although easy to frame as a finger-pointing game, the fraud-detection subject is important because fake plays and AI slop aren’t going away.
On the contrary, Deezer recently identified a staggering 85% of on-platform AI music streams as fraudulent, and Sony Music pulled a whopping 135,000 deepfakes of its artists’ music.
Years after Smith’s scheme began, are DSPs up to the task of preventing AI streaming fraud? Time will tell, but a few warning signs stand out as especially concerning. First, at the top level, Apple Music and others still haven’t booted all Smith’s music; The Best So Far was alive and well at the time of writing.
Then there are questions at the intersection of endless AI “artist” profiles and curious approaches to punishing fraud. Besides the threat of a federal indictment – of which there’s been one thus far – what’s stopping bad actors from gaming the system moving forward? (It seems safe to say that distributor fines haven’t eliminated the problem.)
This ties back to the most worrisome of the warning signs: The apparent lack of coordinated fraud-prevention efforts.
Evidently, even after Spotify, Pandora, and more learned of Smith’s operation, they didn’t give their competitors a heads up or any sort of nudge via DIMA.
Nor did they let the Mechanical Licensing Collective (MLC) – which, incidentally, is embroiled in an ugly legal battle with Spotify – know of their findings. And to put it generously, the MLC didn’t immediately notice and flag the hundreds of thousands of works registered to Smith (and others, but that’s a different can of worms).
All these facts are significant given the scheme’s scope and timing: According to the indictment, Smith in February 2019 signed an agreement to obtain between 1,000 and 10,000 AI tracks per month. And as of 2019, Spotify said it was fielding around 40,000 daily song uploads.
Running with the latter number (which was presumably smaller for rival DSPs) and the high-end distribution total, Smith alone could have been uploading almost 1% of all the new music on streaming services every month. For the platforms that failed to recognize as much, can their since-upgraded fraud-detection capabilities keep up with AI during the remainder of 2026 and beyond?
