Let’s set the scene. A man in North Carolina spends years uploading hundreds of thousands of AI-generated songs to streaming platforms, deploys bots to stream them billions of times, and pockets more than $8 million in royalties that should have gone to real artists. He pleads guilty to conspiracy to commit wire fraud. And then, through his lawyers, he asks the judge for probation.
His argument? No artist suffered any perceptible harm. This is not a parody. It is a filing in Manhattan federal court, case United States v. Michael Smith, No. 24 Cr. 504 (JGK), and it is the most honest encapsulation of how the streaming economy actually works that anyone has yet put in writing — even if that wasn’t quite the intent.
The Argument, Stated Plainly
Michael Smith’s lawyers, in their sentencing memorandum filed September 22, argue that their client’s scheme caused only “tiny, immeasurable losses” spread across millions of songwriters and artists. The specific metaphor they chose, as reported by Billboard, is this:
“The effect each individual stream of one artist’s song has on every other artist’s share of total streams is not akin to a drop in a proverbial bucket — it is a drop in an ocean.”
They are not wrong about the mechanism. They are simply describing it as exculpatory when it is, in fact, the indictment.
The streaming royalty pool is a zero-sum system. Every dollar that flows to a fake stream is a dollar that doesn’t flow to a real one. The reason the harm to any individual artist is “immeasurable” is precisely because Smith spread it across millions of them simultaneously — extracting $8 million from a pool that was supposed to compensate human creators, diluting everyone’s share by a tiny amount, and keeping the difference. The defense is essentially arguing that stealing from a very large number of people in very small amounts is not really stealing.
The lawyers also argued, somewhat adventurously, that “big labels” were doing the same thing. No further elaboration was provided on that point in the publicly available filings.
The Sentence That Hasn’t Happened Yet
Smith pleaded guilty in March to a single count of conspiracy to commit wire fraud, which carries a maximum sentence of five years. Federal sentencing guidelines call for between 46 and 57 months. The US Probation Office’s presentence report recommended 24 months. The defense is asking for probation. Prosecutors have yet to file their own sentencing submission.
Smith was originally due to be sentenced on July 29. That date passed without a sentencing. As of this writing, no new date has appeared on the public docket. As part of his plea deal, Smith agreed to forfeit $8,091,843.64 — which is an admirably precise number for a scheme that his own lawyers are now describing as causing only immeasurable harm. Whatever sentence Judge John G. Koeltl eventually imposes will be the first in United States history for AI-assisted streaming fraud — the first time a federal court has put a number on what this kind of crime is worth in terms of prison time.
The Industry Has Known About This For Years
Here is where the sardonic tone becomes unavoidable. The mechanisms that allowed Michael Smith to extract $8 million from the streaming royalty pool were not secret. They were not sophisticated. They were not novel.
Artificial streaming has been a documented problem since at least 2017, when the major streaming platforms began publishing vague statements about “suspicious activity” in their data. The Mechanical Licensing Collective, to its credit, flagged irregularities in Smith’s streaming data and held back the related mechanical royalties — which is how the scheme was eventually detected. But the broader infrastructure that made the scheme possible — a pro-rata royalty pool, no robust real-time detection, a distribution system that allowed anyone to upload unlimited AI-generated content with minimal friction — remained in place year after year while the industry issued statements about its commitment to artist fairness.
In February 2026, IFPI CEO Victoria Oakley and RIAA CEO Mitch Glazier published an op-ed for Music Business Worldwide describing generative AI as having “industrialized” streaming fraud. They were right. But industrialization doesn’t happen overnight. It happens because the conditions for it exist and persist, and because the people responsible for changing those conditions move slowly enough that others can build factories in the meantime. Smith’s indictment alleged that his bots could generate approximately 661,440 streams per day, worth $1,207,128 in annual royalties. He estimated this himself, apparently without much concern about being caught. That is not the behavior of someone operating in the shadows — it is the behavior of someone who understood the system well enough to know that nobody was watching closely enough to stop him.
“No Perceptible Harm” as Industry Critique
The defense’s “no perceptible harm” argument is legally weak — it misunderstands how restitution works in fraud cases, where the harm to the victim class is the aggregate amount stolen, not the per-victim impact. Judge Koeltl is unlikely to be persuaded that stealing $8 million causes no harm simply because the theft was widely distributed.
But as an accidental critique of the streaming royalty system, it lands with more force than the defense probably intended. The reason individual artists cannot point to a specific dollar amount stolen from them is because the pro-rata pool model doesn’t work that way. It never has. Artists have known since the earliest days of streaming that fake streams dilute their royalties — but because the dilution is diffuse, invisible, and spread across millions of tracks, it has always been structurally difficult to quantify, let alone prosecute.
Smith’s scheme is the first time the US Department of Justice has chosen to prosecute streaming fraud as a criminal matter — and even then, the indictment was narrowed from three charges carrying potential decades in prison to a single count of conspiracy with a five-year maximum. US Attorney Jay Clayton put it plainly when Smith pleaded guilty: “Although the songs and listeners were fake, the millions of dollars Smith stole was real.” That clarity has not always been present in the industry’s own handling of the issue.
What the Sentence Will Actually Settle
The sentencing in United States v. Michael Smith will establish something the music industry has been arguing about in policy documents and op-eds for years: what AI-assisted streaming fraud is actually worth, in terms of criminal consequence, under US law.
If Smith receives a custodial sentence in the range the guidelines suggest — 46 to 57 months — it sends a signal that this category of crime is serious enough to warrant serious punishment, even when the harm to any individual victim is difficult to measure. If he receives probation, or a significantly reduced sentence, the signal is different: that the “drop in an ocean” framing has some legal traction, and that future fraudsters might reasonably expect courts to weigh the diffuse nature of the harm in their favor.
The irony is that Smith’s lawyers have articulated the strongest possible argument for why the streaming royalty system needs fundamental structural reform — and they’ve done it in a document asking a judge to go easy on the man who exploited those structural flaws for $8 million. The sentence hasn’t come yet. When it does, we’ll update this piece. In the meantime, the drop-in-an-ocean defense is sitting in the public record, waiting for the next person who reads it and decides the ocean looks inviting.
Rudy (32) currently based in Bergamo, here since 2019.
https://www.linkedin.com/in/rudy-cassago-522452179/