Every time you tap play, a fraction of a cent changes hands.
You probably never think about where your music subscription goes after it leaves your checking account. Most of it flows into a shared royalty pool that streaming services split among artists and songwriters, based on each song’s share of total plays.
That setup has paid working musicians for well over a decade. Artists get paid, listeners get a near-endless library, and the platforms keep the lights on.
The system rewards volume above everything else, and it assumes every play comes from a person.
That assumption is the weak spot. A bot never gets bored, never skips a track, and never sleeps, and if you feed it enough songs, it can pull money out of a pool meant for real artists.
One man spent seven years proving exactly that. On Oct. 6, the bill arrived.
U.S. District Judge John G. Koeltl sentenced Michael Smith, 54, of Cornelius, North Carolina, to 18 months in federal prison for a scheme that used thousands of bot accounts to stream hundreds of thousands of AI-generated songs, the U.S. Attorney’s Office for the Southern District of New York said.
Smith collected more than $8 million in royalties from Amazon (AMZN) Music, Apple (AAPL) Music, Spotify (SPOT), and Alphabet’s (GOOGL) YouTube Music, prosecutors said. He pleaded guilty in March to one count of conspiracy to commit wire fraud.
“Michael Smith exploited super intelligence technology to generate a fraud,” U.S. Attorney Jamie McDonald said, as reported by WUNC.
Prosecutors called it the Justice Department’s first criminal case of its kind.
In my analysis, the bigger story is how cheaply the scheme scaled and what it shows about where your streaming dollars actually go.
How 10,000 bot accounts turned AI songs into royalty checks
Smith’s playbook ran in three steps, the Justice Department noted. He created thousands of fake streaming accounts, used software to make them play songs he owned around the clock, and collected the royalties those plays generated.
At times he ran as many as 10,000 bot accounts at once, built with fake email addresses and fraudulently obtained debit cards, the department said.
The songs came from AI. Starting around 2018, Smith worked with the CEO of an AI music company who supplied him with thousands of tracks a week, according to WUNC‘s account of the indictment.
Related: Spotify surprises subscribers with a major new free addition
Volume was the whole point. A single song with millions of plays would have tripped the platforms’ fraud alarms, so he spread the plays thin across a giant catalog.
In one email, Smith wrote that he needed “a TON of songs fast” to get around the platforms’ anti-fraud policies, the Justice Department said.
The results got absurd. In April 2023, Taylor Swift’s entire catalog drew 9.3 million family-plan streams on YouTube Music, while Smith’s bots used family plans to stream his AI tracks 80.9 million times that same month, prosecutors said.
That’s nearly nine times Swift’s total, from songs no listener ever picked.
Why every fake song play shrinks a real artist’s paycheck
Here’s the part that matters to you as a listener.
Streaming services pay royalties proportionally from a shared pool of money, so fake plays divert cash from artists whose songs real people streamed, the Justice Department explained.
More Digital Entertainment:
- YouTube just sent creators a message about their views
- Roblox’s hottest new game is already cooling off
- YouTube TV just gave subscribers a reason to look elsewhere
Think of it as a pizza cut by headcount. Every bot that pulls up a chair makes everyone else’s slice thinner, even though the pizza never got any bigger.
Your monthly fee and the platform’s ad revenue fill the pot, and the split depends on each track’s share of all streams in that period, as Music Business Worldwide explained in its coverage of the case.
Smith’s lawyers argued that no recording artist or songwriter suffered any appreciable loss. The music industry pushed back hard.
The International Federation of the Phonographic Industry (IFPI) said the result “sends a clear message that streaming fraud is a crime,” according to Music Week.
The sentence still landed well below what the government wanted. Here’s how the numbers stacked up:
- Prison term: 18 months, plus two years of supervised release
- Forfeiture: $8,091,843.64, the loss figure both sides agreed on
- Prosecutors’ request: At least 46 months, the low end of a 46-to-57-month guidelines range
- U.S. Probation Office recommendation: 24 months
- Defense request: Probation with no prison time
- Total received: More than $14 million from royalty-paying entities, as the government’s sentencing letter noted
- Sources: U.S. Justice Department, Music Business Worldwide
The $8 million forfeiture undercounts Smith’s haul
The headline number undersells the cash flow. The agreed loss subtracts estimated royalties from legitimate plays of Smith’s music, plus money he paid back into royalty pools by buying the accounts his bots used, Music Business Worldwide reported.
His lawyers, for their part, argued that the $8 million figure “far exceeds any amount Mr. Smith actually ‘pocketed,’” according to the same report.
I ran the numbers on the forfeiture alone. Spread across the seven years the scheme ran, $8.09 million works out to about $1.16 million a year, or roughly $3,170 a day.
The pace of the bots is even more telling. One early model Smith emailed himself used 1,040 bot accounts to produce about 661,440 streams a day, Music Business Worldwide reported.
That’s about 636 plays per bot every day, or one song roughly every two minutes and 16 seconds, around the clock. No human listens like that, which is why the scheme needed hundreds of thousands of songs to hide the pattern.
Spotify told Music Business Worldwide in 2024 that its safeguards had limited Smith’s earnings on its platform to about $60,000. Since April 2024, Spotify has paid recording royalties only on tracks with at least 1,000 streams over the previous 12 months.
Deezer data shows machine-made music scams are spreading
Smith’s case looks less like an outlier once you check the industry’s own data.
AI-generated tracks topped 50% of all new daily uploads on Deezer at their June 2026 peak, about 90,000 songs a day, the Paris-based streamer said in a July 21 release.
The same release said up to 85% of streams on fully AI-generated tracks were fraudulent in 2025. Deezer said it demonetizes those streams and will remove AI tracks tied to fraud or left unplayed for six months.
Spotify removed 75 million AI-generated “spam” tracks in a fraud crackdown in September 2025, DJ Mag reported.
AI makes songs nearly free to produce. Bots make plays nearly free to fake. Put the two together, and royalty theft turns into a volume business.
What side hustlers using song generators should know
If you’ve thought about uploading AI-made tracks for extra income, the line in this case is clear.
Smith’s lawyers pointed to a September 2025 Spotify statement that, for royalty purposes, all music is treated equally, regardless of the tools used to make it, Music Business Worldwide reported.
The crime was the fake audience. Here’s how I’d protect your money and your catalog if you’re experimenting:
- Skip any promo service that guarantees plays. Bought streams can look a lot like bot streams to a fraud filter, and that can put your whole catalog at risk.
- Run the math before you spend a dime. On Spotify, a track under 1,000 streams in a year earns no recording royalties at all, so a big catalog of barely played songs can earn you nothing.
- Keep expectations grounded. Deezer says fully AI-generated music makes up just 1% to 3% of total streams on its platform, a sign of how small its real audience is.
Streaming fraud now carries real prison time
Eighteen months is far less than prosecutors wanted. Still, it’s the first U.S. prison sentence for this kind of fraud, and it puts a price on a scheme that once looked like a gray area.
For the industry, the signal is that enforcement has caught up with the tools. For you, it’s a reminder that the money in your subscription is only as fair as the plays it pays for.
Your monthly fee is supposed to be a vote for the artists you love. This case is the first sign that stuffing that ballot box with bots now comes with a cell.
Related: How AI is changing the economics of fraud and cybercrime
