Fraud in the streaming music It has ceased to be a mere suspicion whispered in the hallways and has become a problem with names, figures, and a far-reaching criminal case. What was once suspected as tricks to inflate plays or sneak songs into playlists is now revealed as an industrial system capable of automate both the creation and consumption of music.
The protagonist of this case is Michael SmithA North Carolina producer has pleaded guilty to running a scheme that amassed more than $8 million in royalties through songs generated by artificial intelligence and a massive network of automated accounts. Their scheme not only deceived streaming platforms: He diverted money that belonged to real musicians. worldwide, including those that depend on the European digital market.
The mastermind behind the fraud: AI-created music and nonexistent listeners
According to the documentation of United States Department of JusticeBetween 2017 and 2024, Smith assembled a system that was as simple in concept as it was complex in its execution: mass-produce songs with AI and create a fake audience that I listened to them nonstop on the main streaming services.
Instead of chasing the typical viral hit, he focused on volume. He used generative artificial intelligence software to produce thousands upon thousands of musical pieces, many of them similar to the ambient music that fills relaxation, study, or sleep playlists. Then, I uploaded that catalog to Spotify, Apple Music, Amazon Music, and YouTube Music., usually under the names of unknown artists and obscure labels.
The other half of the plan relied on the fictitious listeners. Smith controlled thousands of automated accountshosted on cloud services and connected from different locations to simulate users spread across the globe. Many of these accounts even had paid subscriptionswhich made the consumption pattern they generated even more credible.
The system operated tirelessly, twenty-four hours a day: the accounts played songs from its catalog in a distributed manner, without focusing on any single track to avoid raising suspicion. In this way, it managed to generate up to 661.000 daily views, spread across hundreds of thousands of tracks, which made it difficult for the platforms' algorithms to detect anomalies at first.
According to the prosecution's calculations, the network allowed him to enter more than [amount missing] in a few years. $1 million annually in royalties, even though no human listener was actually behind it of that enormous volume of listening.
Eight million in royalties and a hole for real artists
The core of the problem lies in the way in which Spotify, Apple Music, and YouTube Music share the moneyMost streaming services use a pooled revenue model: the amounts from subscriptions and advertising are pooled and distributed based on the percentage of plays that each song accumulates within the total.
In this context, artificially inflating the listenership of a fictitious catalog is not a simple numbers game: it implies that Each counterfeit reproduction reduces another artist's income. whose listeners are real people. US prosecutors have repeated this in several appearances: Smith's songs and listeners were invented, but The millions diverted belonged to musicians, composers, and legitimate rights holders..
The figures from the investigation give an idea of ​​the impact: the accused even registered 661.440 daily views and pocketed around 10 million dollars in totalOf which he must return just over 8 million as part of the court agreement. The court has also ordered the confiscation of assets linked to the fraud.
The case has been described by the U.S. Attorney's Office for the Southern District of New York as the first major criminal precedent for music fraud using AIProsecutor Damian Williams himself has stressed that these kinds of schemes "directly steal" money from artists whose songs are actually listened to by real audiences, drawing a red line regarding the manipulation of streaming metrics.
In addition to US platforms, the effect extends across the entire global ecosystem, because The distribution of royalties is calculated on an international scale.When a catalog inflated with bots sneaks into the statistics, creators in other markets—including the European and Spanish ones—see their share of the pie shrink, even if they scrupulously comply with the rules.
How the scam was uncovered: anomalies in listening and institutional coordination
The fraud wasn't discovered by chance. The initial alert came from the Mechanical Licensing Collective (MLC), the entity responsible for managing certain copyrights in the digital environment in the United States, which detected unusual listening patterns linked to a specific group of songs.
These anomalies—very high reproduction volumes, concentrated in little-known catalogs and distributed suspiciously evenly—led the MLC to contact federal authorities. From there, the Complex Fraud and Cybercrime Unit of the Prosecutor's Office and the FBI They reconstructed the network, analyzing server logs, royalty payments, and the bot infrastructure used.
The investigation revealed that Smith had had the collaboration of music technology companiesHe allegedly gave a portion of his monthly profits to these companies in exchange for access to advanced audio generation tools. Although court documents do not name names, specialized media outlets link part of the defendant's catalog to commercial AI-powered music creation platforms.
Finally, faced with the weight of the evidence, Smith pleaded guilty in a federal court in New York to conspiracy to commit wire fraud, in addition to other charges related to the use of bot networks and money laundering associated with illegal income.
The penalty he faces is around five years in prison and the full return of the funds obtained through his scheme. For the US justice system, the message is clear: Manipulating streaming with AI and bots is no longer considered a simple marketing trickbut an economic crime with criminal consequences.
Spotify, Apple Music and the new AI-powered map of music fraud
As the Smith case progressed, the major platforms were already beginning to deal with other episodes of streaming fraudIn 2023, for example, Spotify removed tens of thousands of songs from boomy, an AI-powered music creation application, after detecting an anomalous volume of automated listening intended to inflate the metrics of certain users.
Another relevant incident occurred when a group called Syntax error uploaded software-generated tracks imitating deceased artists, trying to take advantage of the notoriety of their names to capture reproductions and royalties without authorization or control of their heirs or record labels.
These cases have forced services such as Spotify, Apple Music or Deezer to strengthen their integrity and security teams. Among the measures that have been put forward are the implementation of AI-generated specific tags for songs, the finer tracking of listening patterns and the early detection of click farms and fake promotion services.
Platforms like Deezer have warned that new tracks are being added to their catalogs every day. tens of thousands of tracks created entirely by AIIt is estimated that a single advanced tool can generate up to seven million songs daily, an amount that would allow the complete catalog of a standard streaming service to be recreated in just a couple of weeks.
The consequence is obvious: faced with an ocean of synthetic content, control systems have a harder time. distinguish between legitimate activity and organized fraudAnd human artists are forced to compete for visibility and income with automated catalogs whose production cost is almost zero.
Europe and Spain: a global problem that is already appearing at home
Although the case was tried in the United States, its repercussions directly affect Spain and the European marketThe platforms' business model is global, and the massive manipulation of views in one country ends up affecting the international royalty distribution calculations, with an impact on EU member states.
In recent years, the European Union has tightened the obligations of large digital platforms through regulations such as Digital Services Act (DSA)which increases its responsibility in detecting and eliminating illicit activities. Added to this is the imminent European AI regulationwhich will require transparency and additional controls over systems that generate or process automated content.
Meanwhile, collective management organizations and industry associations in countries like Spain are demanding better mechanisms for tracing the origin of reproductionsas well as clearly identifying which works have been created by artificial intelligence. The idea is not to ban AI, but to prevent it from becoming a means to to quietly divert income from artists who rely on streaming.
For many Spanish musicians, especially independent ones, the concern is twofold. On the one hand, they fear the unfair competition from mass-produced catalogs These projects occupy space in background playlists and capture small portions of the overall payment pool. On the other hand, they are concerned that, faced with increased fraud, platforms will respond by tightening their monetization criteria, which could make it even more difficult for emerging projects to earn significant amounts.
In this context, the Smith case serves as a warning: if the anti-fraud systems and regulationThe current royalty distribution model risks becoming unsustainable for those who make a living from music, both in the United States and in Europe.
Industrial-scale musical AI: the perfect breeding ground for fraud
The scandal has coincided with the popularization of services such as Suno and other automatic composition tools, which allow anyone to generate complete tracks—with vocals, lyrics, and arrangements—in a matter of seconds. These applications have democratized music production, but they have also created a ideal environment for abuse.
Deezer and other industry players estimate that daily uploads to the platforms tens of thousands of songs generated solely by AISome research cited by specialized media points out that the potential volume reaches figures of millions of tracks per day, which overwhelms the capacity of manual review and puts detection algorithms in a difficult position.
The boundary between human and automated systems is becoming increasingly blurred: several studies indicate that around 97% of listeners are unable to distinguish whether a song was composed by a person or by an AI system. This makes synthetic music possible blend seamlessly into playlists alongside traditional productions, competing on equal terms for the same reproductions.
Even within the tech industry itself, there are doubts about the direction the sector is taking. Those in charge of music AI platforms have publicly acknowledged that they live with a certain "permanent indecision," aware that the flood of automated content It can erode the chances of new generations of artists building a sustained career over time.
In this scenario, the combination of music generated on an industrial scale and bot farms This becomes a particularly dangerous mix: the volume of tracks allows for the concealment of false listening patterns, while the pay-per-play system makes it easy to monetize even very small percentages of the common royalty fund.
Market reaction: anti-fraud algorithms, data, and new payment models
The legal precedent set by the Michael Smith case has accelerated the industry's response. Major platforms and many record labels are investing in advanced data analysis systems to detect anomalies earlier and stop similar schemes before they reach millions of dollars.
Among the solutions being explored are limits on the amount of content that a single user can upload In a specific period, stricter verification processes for certain accounts and manual reviews when an unknown catalog begins to generate disproportionate income in a very short time.
This technological effort also opens up a new business niche. Proposals are beginning to emerge for SaaS platforms specializing in proactive anomaly detection in streams, payments, and metadata, designed for aggregators, labels, and streaming services. In the EU, where regulatory compliance is especially important, these types of tools could become a standard requirement for operating smoothly.
At the same time, they discuss alternative income sharing modelsOne idea gaining traction is the user-centric system, in which the money each user pays is distributed only among the artists that listener has listened to, instead of going into a global fund. While it doesn't completely eliminate the problem of bot farms, it could reduce some incentives to create massive catalogs designed to scrape together small fractions of the total.
The big challenge for the sector is to find a point of equilibrium where the AI is used as a creative tool without becoming a weapon to dismantle the remuneration system. Without solid control mechanisms, confidence in the streaming model—already heavily criticized for low unit payments—could be further damaged, with consequences for artists, labels, and listeners.
The case of the North Carolina producer has starkly illustrated how the combination of AI-generated music, bots, and a pay-per-play model It can siphon off millions without anyone actually listening to those songs. The response from the US justice system, the regulatory movement in Europe, and the technical reaction of platforms like Spotify, Apple Music, and YouTube Music will determine whether streaming remains a viable space for creators or becomes increasingly dominated by phantom catalogs that no one knows about, but which generate revenue as if they were genuine hits.
