Ai-generated songs rake in $8 million before feds mute the stream
Michael Smith’s playlist was infinite, his listener base fake, and his royalty checks obscene: 661,440 plays a day, every day, funneled through bot accounts he built to impersonate human ears. The 52-year-old North Carolina operator netted more than $8 million before investigators traced the static back to him, culminating in a guilty plea for wire-fraud conspiracy that carries up to five years in federal prison.
How one man gamed spotify, apple and amazon at once
Smith’s formula was deceptively simple. First, he commissioned thousands of tracks from AI composition tools, slapping random-artist names on files no human ever sang. Next, he spun up a cloud army—roughly 10,000 bot accounts across Spotify, Apple Music, Amazon Music and YouTube Music—programmed to stream those tracks on loop, 24/7. The platforms’ pro-rata payout models did the rest, converting each artificial play into fractions of a cent that snowballed into seven-figure annual revenues.
At peak, Smith’s network delivered more than 600,000 daily streams, a volume that would require a mid-tier indie act with genuine fans. Royalty statements showed yearly income topping $1.2 million, money he then shuffled through shell companies and prepaid debit cards to hide the trail. Fraud-detection teams at multiple distributors finally spotted the anomaly: identical listening patterns, zero playlist adds, and IP clusters that looked like data-center maps rather than bedroom Wi-Fi.
The precedent prosecutors wanted to set
Federal charging papers describe the scheme as the largest AI-driven music scam prosecuted anywhere. Smith faced three counts carrying up to 20 years each; the government asked for a sentence that would have kept him locked up until age 90. A plea deal trimmed exposure to a single conspiracy count, capping incarceration at five years. Sentencing is set for 29 July in Charlotte, where attorneys will argue whether the punishment deters copycats—or merely teaches them to code smarter.
Investigators note that Smith diversified across platforms precisely to avoid tripping internal alarms, yet the uniformity of his bot behavior became a fingerprint. Denmark saw a smaller version of the same ruse last year; industry lawyers warn streaming fraud has mutated from lone-wolf experiments into organized, AI-accelerated enterprises that can mint millions before detection algorithms adapt.
The platforms insist they have since patched the loopholes, but royalty audits remain opaque. For every Smith caught, labels privately admit there are quieter operators still racking up plays, one algorithmic beat at a time. The message from prosecutors is blunt: code may write the songs, but humans still do the time.