He Outstreamed Taylor Swift With 10,000 Bots. Now He's Going to Prison.

In April 2023, one artist's catalog racked up 80.9 million streams on YouTube Music family plans in a single month. That artist was not Taylor Swift. Her entire catalog pulled 9.3 million family-plan streams that same month, barely a tenth of the mystery catalog's total. The mystery artist was Michael Smith, a 54-year-old musician from Cornelius, North Carolina, and his fans were not real. They were bots.

On October 6, 2026, Smith was sentenced to 18 months in federal prison for the scheme, the first U.S. criminal case over AI-assisted music streaming fraud. U.S. District Judge John G. Koeltl, in a Manhattan courtroom, also ordered two years of supervised release and the forfeiture of $8,091,843.64, every dollar the government calculated he had extracted from the royalty pools that pay real musicians.

A small empire of fake listeners

The operation ran from 2017 through 2024. Prosecutors say Smith used fake email addresses and fraudulently obtained debit cards to create thousands of accounts on Spotify, Apple Music, Amazon Music, and YouTube Music, then ran software that kept those accounts streaming his music around the clock. At peak, as many as 10,000 bot accounts were active at once, many running on cloud computing, many signed up for the cheapest family plans.

The family-plan detail mattered. Multiple listeners under one subscription made industrial-scale fraud look like heavy household use. Investigators say the manipulation exploited exactly that structure, blending in with ordinary family listening instead of standing out as a single account playing the same song 10,000 times.

Then came the second innovation: volume of material. As the operation grew, Smith turned to artificial intelligence to generate hundreds of thousands of songs, spreading the fraudulent plays across a vast catalog so no single track's numbers looked suspicious. In an email cited by the government, he told a collaborator he needed "a TON of songs fast" to work around streaming platforms' anti-fraud policies. AI could compose faster than fraud teams could investigate.

His fake audience outstreamed Taylor Swift nearly nine to one. Every one of those plays took money out of the pool that pays real musicians.

The math of a fraud

Smith did his own accounting. In October 2017 he calculated that the operation was generating 661,440 streams a day, a pace worth $1,207,128 a year. That works out to roughly half a cent per fake play, which explains the industrial scale: getting to $8 million at half a cent a stream takes billions of plays, around 241 million a year by his own numbers.

The streams were real in one sense. They happened, and they consumed real server capacity. What they were not was listening. Almost none of the billions of plays represented a human choosing to hear a song. The money, though, was real, and the way streaming royalties work made the fraud possible. Services distribute royalties from pools of revenue according to each catalog's share of plays. Artificially inflating one catalog's share redirects money toward the fraudster and away from every other rights holder in the pool.

Caught in 2018, paid until 2024

The most uncomfortable part of the story is how early the scheme was detected. A music distributor flagged Smith's streams as abuse in March 2018 and warned that October it would pull his releases. Smith called the warning "absolutely wrong and crazy," according to reporting on the case. The streaming services kept paying.

For six more years, the bots kept streaming, the royalties kept flowing, and the operation kept growing. Federal prosecutors didn't charge Smith until September 2024, with a three-count indictment covering wire fraud conspiracy, wire fraud, and money laundering conspiracy, each carrying a potential 20-year sentence. In March 2026 he pleaded guilty to a single count of conspiracy to commit wire fraud, which carries a five-year maximum.

The delay matters. Between the first abuse flag and the indictment, Smith collected the bulk of the $8 million. Detection systems that identify fraud but don't cut off payments are an interesting kind of failure, and it's one the industry is now being forced to reckon with.

The sentencing

Federal sentencing guidelines called for 46 to 57 months in prison, and prosecutors asked the judge to impose at least 46. The probation office recommended two years. Smith's lawyers argued for probation with no prison time, saying the financial harm to any individual artist was tiny because the losses were spread across millions of rights holders. They compared the impact of each fraudulent stream to a drop in the ocean.

Prosecutors rejected that framing. The distributed nature of the losses, they argued, didn't change the fact that $8,091,843.64 flowed to Smith. "By flooding music streaming platforms with automated bots in the place of consumers, and fake songs in the place of creativity, Smith robbed millions in royalty payments from genuine artists and their fans," U.S. Attorney Jamie McDonald said. Judge Koeltl ultimately imposed 18 months, below what the government wanted but still prison time for a case federal prosecutors have described as the first of its kind.

Why this case changes music

One important boundary: the case did not treat AI-generated music itself as criminal. Plenty of legitimate music is made with AI tools. Prosecutors focused instead on the fraudulent accounts and automated streams, on creating the appearance of genuine listeners to obtain royalty payments for activity that never happened. Making music with machines is legal. Inventing an audience of 10,000 fake people to get paid for it is wire fraud.

The industry response has been swift. The Recording Industry Association of America said after the sentencing that streaming fraud harms creators by diverting money generated by legitimate listeners. The Music Fights Fraud Alliance called the prosecution an important deterrent while warning that broader efforts will be needed. Labels and platforms say the case will accelerate investment in detecting AI-generated content and bot activity.

The structural problem remains. Streaming pays per play from a fixed pool, which means any successful fraud is a tax on every honest artist. As long as bots can be rented cheaply and songs can be generated in bulk, the economic incentive for this kind of operation persists. Smith's case proves it can be prosecuted. Whether it can be prevented is a harder question, and the answer will determine how much of streaming's payout pool actually reaches the musicians it was meant for.