The $8 Million AI Music Scam: How One Man Used AI Songs and Bot Traffic to Defraud Streaming Platforms

How a Former Healthcare Operator Used AI Music to Pocket $8 Million

According to a report on The Vergecast podcast, a case that Wired journalist Kate Nibbs spent years tracking has finally come to a close: Mike Smith, a businessman who once ran a chain of urgent care clinics in North Carolina, was sentenced to 18 months in federal prison and ordered to repay $8,091,843.64 in illicit proceeds for using AI-generated music and bot-driven streaming to commit fraud. It marks the first criminal prosecution in the United States targeting AI streaming fraud.

Smith didn't arrive at this path overnight. In his earlier years, he self-funded music videos, signed artists, and even fully bankrolled a rap talent competition reality show that aired on BET. Kate Nibbs described him on the podcast as "an ordinary guy who can play guitar" — his work was hardly outstanding, but he had spent years grinding at the edges of the music industry without ever truly finding an audience.

This background is crucial — Smith's behavior is closer to that of a "shortcut addict" than a mere opportunistic arbitrageur. He kept pouring profits from the healthcare industry into his music career, trying virtually every avenue imaginable to become famous, until AI tools gave him unprecedented leverage.

How AI Music Generation Tools Became the Engine of Fraud

The core tool Smith used was an AI music generation service called Boomy. Before Suno and Udio entered the public eye, Boomy already allowed users to quickly generate songs from simple prompts. Smith's workflow went roughly as follows:

• Generate hundreds of thousands of AI songs in bulk using Boomy
• Create a large number of fictitious artist identities
• Upload the songs to mainstream streaming platforms through distribution services like DistroKid
• Build a bot network to loop-play these songs and inflate streaming numbers
• Route the royalty revenue back to himself

Kate Nibbs pointed out that publishing AI music on streaming platforms is not itself illegal. Platforms like Spotify never explicitly banned AI-generated content — Spotify even eventually removed 75 million generated songs in a single sweep, but that was a platform governance action, not a legal requirement. What actually got Smith convicted was streaming fraud: using bots to inflate play counts violated platform terms of service, and his repeated denials when questioned ultimately led prosecutors to characterize his conduct as fraud.

Anti-fraud company executive Morgan Hayduk told Nibbs that Smith made one critical mistake: he operated entirely from within the United States. Many similar AI arbitrageurs choose to operate remotely from places like Cambodia, which greatly increases the difficulty of enforcement.

Boomy is an AI music generation service aimed at ordinary users. No knowledge of music theory is required — users simply select style tags or enter a short prompt, and a complete song is generated within seconds. It entered the market earlier than Suno and Udio, initially marketed itself on the promise that "anyone can release a song," and once partnered with streaming distribution channels to help users push their work directly to Spotify, Apple Music, and other platforms. This low-barrier, batch-oriented nature makes it naturally prone to abuse — a single user can generate hundreds or thousands of stylistically similar tracks in an extremely short time, and in the early days platforms did not set strict upload limits or identity verification for generated content, creating fertile ground for subsequent streaming fraud.

Why Didn't He Just Inflate His Own Songs?

Podcast host David Pierce raised a thought-provoking question: if Smith had only used bots to inflate his own streams and package himself as a successful musician, wouldn't he have been harder to catch?

Nibbs explained that a sudden spike in play counts for a single artist would immediately trigger platform risk-control systems. Smith chose to spread bot traffic across hundreds of thousands of AI-generated fictitious artists, which is logically harder to detect. This is also the general strategy behind today's flood of AI spam — dilute risk through sheer volume.

But Nibbs also noted that this strategy has a fatal weakness: when your songs chart on Billboard but you have no real fans, industry insiders will get suspicious. Her conversation with Smith's former business partner, Jonathan Hay, confirmed this — Hay had noticed these strange streaming numbers long ago.

Streaming platform risk-control systems typically monitor abnormal play curves for individual tracks or accounts. If an artist's play count suddenly spikes unnaturally, with listening activity concentrated among the same batch of IPs or devices, the system triggers manual review or even direct removal. Smith spread bot traffic across hundreds of thousands of fictitious artists precisely to dilute the anomaly so that each account appeared "normal." This strategy is known in the AI spam world as "volume cover" — drowning detection thresholds with massive amounts of low-quality content so platforms can't screen them one by one. But the cost is that individual accounts struggle to accumulate real fans, and once a fictitious identity unexpectedly lands on a chart like Billboard, the lack of supporting evidence such as live performances, social media engagement, and media coverage will instead draw the attention of industry insiders.

What This Case Means for the Entire AI Content Ecosystem

Smith's sentencing comes against the backdrop of explosive growth in AI-generated content. Nibbs began reporting on the AI spam problem in 2023, and she observed a clear shift in platform responses: initially Amazon was almost oblivious to the flood of AI-generated books, but now it has begun limiting self-publishing volumes; LinkedIn has also introduced a report button for AI content.

But are these measures enough? Nibbs is skeptical. She believes AI arbitrageurs seeing the Smith case won't respond by stopping — they'll respond by saying, "I'll operate from Thailand, get paid in crypto, and hide my identity."

A more fundamental question: as AI-assisted creation becomes the norm, how do platforms draw the line between "AI-generated" and "AI-assisted"? Nibbs admits that purely AI-generated songs (like the output of Suno and Udio) may technically be detectable, but music made with AI involvement in the production process will become increasingly difficult to distinguish.

Pierce also pointed out on the show that platforms like TikTok have almost no incentive to crack down on purchased fake streams — because it neither directly harms platform revenue nor degrades user experience to the point of requiring action. Spotify went after Smith not because he uploaded AI music, but because bot-driven streaming directly eroded the royalty pool that should have been allocated to real artists.

In other words, governance only truly kicks in when fraudulent behavior directly harms the economic interests of the platform or rights holders.

The royalty pool is the mechanism by which streaming platforms distribute total revenue to rights holders in proportion to total play counts. When bots rack up massive fake streams, those plays are counted in the total denominator, diluting the share of real artists. The reason platforms like Spotify actively pursued Smith is not AI music itself, but that streaming fraud directly eroded income that should have gone to real rights holders. This also explains why platform incentives to crack down on purchased fake streams vary so widely: if fake traffic doesn't noticeably affect a platform's own advertising revenue or user retention, the platform often lacks the economic incentive to enforce proactively — only when the interests of downstream rights holders are harmed to a certain degree does governance actually get pushed forward.

This Isn't Just One Man's Story

Smith ultimately pleaded guilty and faces 18 months in prison. Prosecutors made clear they wanted to make an example of him. But both Nibbs and Pierce believe there are far savvier imitators out there — and this case won't stop them.

From the deluge of AI-generated books on Amazon, to podcast companies mass-producing AI podcast episodes, to prediction markets now being used to manipulate streaming data to influence trading outcomes — the AI arbitrage playbook is spreading to every corner of the internet. The real significance of the Smith case may not be who it punished, but the deeper problem it exposed: when the marginal cost of generated content approaches zero and the decentralized royalty distribution system is extraordinarily complex, the soil for this kind of fraud is far more fertile than we imagine.

As Nibbs put it at the end of the show: "Every morning I wake up grateful that I'm not the person in charge of integrity at a social platform."

One-line Summary

First US criminal case on AI streaming fraud: a former healthcare operator used AI to mass-generate songs and inflate streams, netting $8 million and a prison sentence.

Paragraph Summary

The article recounts the case of Mike Smith, who used Boomy to mass-generate AI songs, created hundreds of thousands of fictitious artists, and inflated streams via bots — ultimately sentenced to 18 months in prison and ordered to repay over $8 million for fraud. It notes that uploading AI music is not itself illegal; the conviction hinged on streaming fraud and his denials when questioned. Smith spread traffic to evade risk controls but was exposed when he charted on Billboard with no real fans. The case unfolded amid the AI content explosion, with platform governance driven mainly by economic harm. The author sees the ruling as more symbolic deterrence, as savvier imitators may already have shifted to cross-border operations and crypto, and AI arbitrage continues to spread.