On October 6, 2026, a federal judge sentenced Michael Smith to 18 months in prison for an AI-assisted streaming fraud scheme that pulled in more than $8 million in royalties. The crime was not making music with AI. It was faking listeners with bots and fake accounts to steal from a shared royalty pool.
That distinction is everything. If you make AI music, you need to understand it cold, because this case just drew the brightest line yet between what is legal and what can land you in federal court.
Smith, 54, of Cornelius, North Carolina, pleaded guilty on March 19, 2026 to one count of conspiracy to commit wire fraud. U.S. District Judge John G. Koeltl, in the Southern District of New York, sentenced him to 18 months in prison, followed by 2 years of supervised release, plus forfeiture of $8,091,843.64. That last number is forfeiture — money the government says came from the crime — not a fine. Prosecutors had urged at least 46 months. Probation had recommended 24 months. The judge went lower than both, at 18 months.
This is not legal advice. I am not your lawyer. It is a plain-English breakdown of what happened, what the government said it proved, and what it means for honest creators who use AI as a tool.
What Actually Happened
The Department of Justice called this the first criminal prosecution involving AI-assisted music streaming fraud. That framing matters. Prosecutors did not charge Smith for using AI to make songs. They charged him for building a machine to fake listening at scale, then collecting real royalty money for fake plays.
The scheme ran from 2017 to 2024, according to prosecutors. Over those years, Smith allegedly created hundreds of thousands of AI-generated songs. On their own, those songs were not the crime. The crime was how he got them played — or, more accurately, how he made it look like they were played.
He used up to roughly 10,000 bot and fake accounts at once across Spotify, Apple Music, Amazon Music, and YouTube Music. Those accounts streamed his tracks billions of times in total. And he spread the plays thinly across a huge catalog, so no single song spiked in a way that would trip an obvious alarm.
Think about that for a second. If you have one song with a million fake plays, someone notices. If you have hundreds of thousands of songs each picking up a modest number of fake plays, the fraud hides in the noise. That was the design.
Prosecutors put a number on the scale that still stops me when I read it. In April 2023 alone, they said, Smith’s bots generated 80.9 million YouTube Music family-plan streams, compared with 9.3 million streams for Taylor Swift’s entire catalogue that same month. One man’s bot network, in one month, on one service, out-streamed one of the biggest artists on the planet by nearly nine to one.
That is not a gray area. That is industrial-scale fakery.
How the Scheme Worked
You do not need to be technical to get the mechanics. The pieces were simple, and that is what made them dangerous in combination.
First, volume. Smith’s operation generated a massive catalog of AI-created tracks. Hundreds of thousands of songs means you never run out of fresh inventory to point bots at, and you never depend on any one track breaking through.
Second, fake listeners. The operation ran as many as about 10,000 bot or fake accounts at a time. Those accounts were not real fans. They did not choose music because they liked it. They played what they were told to play, on repeat, at scale.
Third, distribution and collection. By spreading plays across many tracks and across multiple platforms — Spotify, Apple Music, Amazon Music, YouTube Music — the operation avoided putting all its weight in one place. Streaming services pay royalties out of a shared pool, so when fake streams claim a slice, real artists get less.
U.S. Attorney Jamie McDonald put it this way: “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.”
Smith pleaded guilty to one count of conspiracy to commit wire fraud, which carries a maximum of 5 years in prison.
Why It Mattered: The Shared Pool Problem
Streaming royalties do not come from a bottomless pot that grows every time someone presses play. In the dominant model, services put revenue into a pool, then split that pool based on share of total streams. So a fake stream does not just add a little extra money for the fraudster. It shrinks everyone else’s slice.
That is why prosecutors, artists, and platforms describe streaming fraud as theft, not as a victimless shortcut. Over billions of fake plays, millions of dollars moved from honest work to dishonest automation.
You can feel how personal that is if you have ever released a song yourself. You know what it takes to get 1,000 real plays. Now imagine someone skipping all of that with 10,000 fake accounts and collecting $8 million for it. That is the anger behind this case, and it is justified.
The Sentence: What the Judge Decided
Smith was sentenced on October 6, 2026 to:
- 18 months in federal prison
- 2 years of supervised release after prison
- Forfeiture of $8,091,843.64
Again, that $8,091,843.64 is forfeiture, not a fine. Forfeiture means giving up proceeds tied to the crime. A fine would be a separate penalty.
Prosecutors asked for at least 46 months. Probation recommended 24 months. The judge imposed 18 months. That gap is normal in federal sentencing. Judges weigh the guidelines, the plea, the defendant’s history, and the scope of harm, and they do not have to follow either side’s number.
Smith pleaded guilty on March 19, 2026 to one count of conspiracy to commit wire fraud. That single count carries a maximum of 5 years. The plea meant the government did not have to prove the case at trial.
What This Case Is NOT About
This is the part that gets lost fastest.
This case is not about AI music being illegal. It is not about AI-generated songs being banned. It is not about a court deciding that a song made with AI tools cannot earn royalties.
The crime was bots plus fake accounts plus wire fraud. Take the bots away and you do not have this case. Take the fake accounts away and you do not have this case. The AI-generated songs were the inventory. The bots were the crime.
That matters because a lot of honest creators use AI as part of their process right now. They write their own lyrics. They use an AI vocal tool. They produce, mix, and release real music to real listeners. Nothing in Smith’s sentencing makes that illegal.
I have seen the fear in creator groups already: “Does this mean my AI songs are at risk?” No. Risk comes from how you get plays, not from what tool helped you make the track.
The Bright Line: Lawful AI Music vs. Fraud
Here is the bright line, in plain language you can check your own work against.
Lawful AI music looks like this:
- You write the lyrics yourself, or you have a clear right to use them. You do not copy someone else’s words and pass them off as yours.
- You use AI as a production tool — for vocals, for instrumentation, for arrangement help — the way earlier generations used drum machines, samplers, and auto-tune.
- You disclose AI use where the platform or distributor asks you to. If Suno, your distributor, or the streaming service has a disclosure box, you check it honestly.
- You release through a legitimate distributor and you follow that distributor’s rules on AI content and Content ID.
- Real people find your music, choose to play it, save it, share it, and come back. Growth may be slow. That is normal. Real growth is slow.
- You can explain, if anyone ever asks, exactly how you made the track and exactly how you promoted it, without a single step you would be embarrassed to describe.
Fraud looks like this:
- Plays come from accounts that are not real listeners, or from services that promise plays you did not earn.
- You pay for streams, or for placement that guarantees streams, rather than for honest promotion to real people.
- You run, rent, or benefit from bots, click farms, or fake-account networks.
- You spread fake plays across many tracks to hide the pattern.
- You cannot explain your growth without admitting a step that breaks the platform’s rules.
That is the test. It is not about whether AI touched your song. It is about whether real humans chose to listen.
If you want a deeper breakdown of how distributors treat AI music right now — including Content ID rules that trip up a lot of new creators — read our guide on AI music distributor rules and Content ID. The rules vary by distributor and by tool, so check the current policy for the service you actually use before you upload.
And if you are also using music in short-form video, the copyright rules are different again. Our breakdown on copyright-safe music for monetized reels walks through what you can and cannot use when the video itself is earning money.
The Never-Do List
Print this section in your head. Tattoo it on your release checklist. These are the moves that put creators on the wrong side of the line, even when someone selling a “promo package” tells you everyone does it.
Never buy streams. Not from a website, not from a DM, not from a “promo guy” on Instagram. If someone sells you plays, those plays are either bots or incentivized clicks. Either way, they are fake.
Never use bot farms. Never run bots yourself. Never let a “manager” or “label service” run them for you and tell you not to worry about it. You are the one whose account and catalog are on the line.
Never use playlist stuffing. Paying to get added to a playlist that exists only to farm streams, or stuffing your tracks into low-quality playlists that promise plays, is the same fraud in a nicer outfit.
Never use stream-boosting services. If a service promises a specific number of plays for a price, walk away. Real promotion promises exposure to real people. It never promises a play count.
Never create or use fake accounts to stream your own music. Not ten accounts. Not one account on repeat. Not your cousins’ accounts all looping your song overnight. If the listener is not real and the listening is not genuine, it does not count — and it can count against you.
Never fall for guaranteed-playlist promises. No legitimate playlist curator can guarantee you a set number of streams. A guarantee is a red flag that the plays are not organic.
If you are comparing distributors and wondering which one fits AI music best, do the comparison on their actual current rules, not on a seller’s promise. Our DistroKid vs. Ditto comparison for AI music lays out the differences that matter for AI creators. If you go with DistroKid, you can use this referral link: DistroKid (affiliate link) — it supports this site at no extra cost to you, and you still need to follow DistroKid’s AI and Content ID rules on every release.
One more thing about that never-do list. You do not get to outsource the blame. If a promoter buys streams for you, platforms and prosecutors will still look at your account, your catalog, and your royalties. “My promoter did it” is not a defense. Vet anyone who touches your promotion, get their methods in writing, and cut ties the moment they cannot explain exactly where your plays come from.
What Comes Next: The Copyright Office Inquiry
The day after Smith’s sentencing, the story got a second chapter. On October 7, 2026, the U.S. Copyright Office opened a Notice of Inquiry, Docket No. 2026-6, titled “Music Streaming Fraud.”
This is important, and it is also easy to misunderstand. A Notice of Inquiry is not a new law. It is not a new ban. It is not a rule that takes effect tomorrow. It is a formal request for information — the government saying, “We need to understand this problem better before we decide what, if anything, to do about it.”
The inquiry was prompted by a May 21, 2026 letter from Rep. Scott Fitzgerald. In that letter, Fitzgerald wrote that there is “nothing inherently wrong with using AI as a creative tool in music development” but that generative AI “can serve as a catalyst for criminals to perpetrate streaming fraud.”
That sentence is worth pausing on, because it is the same bright line as the Smith case. AI as a creative tool: fine. AI as a catalyst for fraud: the problem. The congressman and the prosecutors are saying the same thing from different directions.
The Copyright Office inquiry asks 10 sets of questions. In broad terms, it wants to know:
- How do streaming fraud schemes actually work, step by step?
- What do they cost — to artists, to platforms, to the system as a whole?
- How are fraud schemes detected right now, and what works?
- What role does AI-generated music play in fraud, versus its legitimate creative uses?
- What solutions make sense — voluntary industry steps, statutory changes, regulatory action, or some mix?
- Should the royalty model itself change — for example, toward user-centric or artist-centric models that pay based on what each real listener actually played, rather than on a single shared pool?
That last question is a big one. Under a user-centric model, your subscription money goes to the artists you actually listened to, not into a giant pool split by total share. Supporters say that model makes fraud harder and pay fairer. Critics say it is complex to run. The inquiry does not pick a side. It asks for evidence.
Here are the deadlines, exactly as announced:
- Initial comments due: November 23, 2026 at 11:59 PM ET
- Reply comments due: December 21, 2026 at 11:59 PM ET
- Submissions via: regulations.gov
If you are an artist, a distributor, a platform, or just a concerned listener, those are the windows to make your voice heard. A comment grounded in your own experience is far more useful than a generic form letter.
Fitzgerald’s letter also cited a striking data point from Deezer. Deezer has claimed that up to 85% of streams on AI-generated tracks were fraudulent in 2025. Treat that number carefully. It is Deezer’s own claim, not independently verified. It tells you Deezer believes fraud is heavily concentrated in a slice of AI-generated catalog — it does not prove that 85% of all AI music streams everywhere are fake.
For more on the inquiry itself, you can read the reporting from TechRepublic on the sentencing and Digital Music News on the Copyright Office investigation.
What This Means for You If You Make AI Music
Let us bring this home. You are not Michael Smith. You do not have 10,000 fake accounts. You are trying to make good music with new tools and get it heard. What should you actually do differently after this case?
First, keep making music. Do not let a fraud case scare you out of a legitimate creative lane. The government, the Copyright Office, and even the congressman who prompted the inquiry all draw the same line: AI as a tool is fine, fraud is not.
Second, audit your promotion. If you have ever paid for a promo package you did not fully understand, go back and look at what it actually did. Did it promise plays? Did it promise playlist adds with guaranteed numbers? If anything looks off, stop using that service today.
Third, keep records and disclose honestly. Save your lyrics, your session notes, your distributor receipts, and your promotion invoices. If a distributor or platform asks whether a track uses AI, answer truthfully. Hiding AI use creates a second problem — a policy violation — on top of whatever prompted the question.
Fourth, build for real fans, not vanity metrics. A hundred real listeners who save your song and come back next week are worth more than ten thousand fake plays that vanish when the bots get banned. Real fans share your work and stick around. Fake plays just rent you a number that can get your catalog removed.
Finally, watch the Copyright Office process. The inquiry will not change the law overnight, but the comments it collects will shape what Congress and the Copyright Office do next.
How to Stay on the Right Side: A Quick Checklist
Before you release your next AI-assisted track, run through this list. If you can answer yes to every item, you are in good shape.
- Did you write or properly license the lyrics and any sampled material?
- Did you disclose AI use where your distributor or platform requires it?
- Are you releasing through a legitimate distributor whose AI rules you have actually read?
- Is your promotion aimed at real people — social posts, real playlists, real communities, real ads — with no promised play count?
- Can you name, in one sentence, where each stream is supposed to come from?
- Have you said no to every bought-stream, bot, and guaranteed-playlist offer, even the ones that sound professional?
If any answer is no, fix it before you hit publish. It is always cheaper to delay a release than to clean up a policy strike, a catalog takedown, or a frozen royalty account.
And if someone in your circle is still buying streams, send them this article. Not to shame them. To warn them. The Smith case proves the government is not only banning accounts anymore. It is bringing criminal cases.
Frequently Asked Questions
Is making music with AI illegal after the Michael Smith sentencing?
No. Smith was sentenced for conspiracy to commit wire fraud built on bots and fake accounts, not for using AI to make music. Using AI as a creative tool — for example, writing your own lyrics and using an AI vocal tool, with disclosure where required and real listeners — remains lawful. The crime was faking listeners to divert royalty money.
What exactly did Michael Smith do?
Prosecutors said he ran a scheme from 2017 to 2024 that used hundreds of thousands of AI-generated songs and up to about 10,000 bot or fake accounts at once across Spotify, Apple Music, Amazon Music, and YouTube Music. The bots generated billions of streams, spread thinly across tracks to avoid detection, and collected royalties from a shared pool that would otherwise have gone to real artists.
Was the $8,091,843.64 a fine?
No. That amount was forfeiture — proceeds tied to the crime that the court ordered Smith to give up. It was part of a sentence that also included 18 months in federal prison and 2 years of supervised release. Do not describe it as a fine, because a fine is a different kind of penalty.
Does the Copyright Office inquiry mean AI music is about to be banned?
No. The Copyright Office opened a Notice of Inquiry, Docket No. 2026-6, on October 7, 2026 to gather facts about music streaming fraud. An inquiry is fact-finding, not a new law or ban. It asks 10 sets of questions about how fraud works, what it costs, how it is detected, what role AI plays, and what solutions might help. Initial comments are due November 23, 2026 at 11:59 PM ET, with reply comments due December 21, 2026 at 11:59 PM ET via regulations.gov.
How can I tell if a promotion service is selling fake streams?
Watch for the red flags: a promised number of plays for a price, guaranteed playlist placement with guaranteed stream counts, vague answers about where listeners come from, sudden spikes with no saves, follows, or shares, and pressure to pay fast without a clear method. Real promotion promises exposure to real people. It never promises a play count. If a service cannot explain its method in plain language, do not use it.
What should I do if I previously bought streams or used a shady promo service?
Stop using the service immediately. Do not buy more plays to cover the gap. Keep your receipts and any messages that show what the service promised, and do not delete evidence. Review your distributor’s and platforms’ policies, and if you are worried about your account, talk to a qualified attorney before you contact a platform.