Digital Age Scam: $10 Million Raised Through Melodies Composed by Artificial Intelligence Algorithms and Listened to by Fictitious Listeners

A US federal court is trying an online music fraudster who automated, with fake accounts, the listening of his own songs registered on major international platforms. In order to collect royalties.

“When the music is good” sang Jean-Jacques Goldman in 1982. This is not the opinion of the judges of the Federal Court of the State of New York who published on September 4 the charges against a certain Michael Smith.

He is accused of fraudulently collecting more than $10 million (€9 million) in royalties between 2017 and 2024, thanks to his musical publications on major sites such as Amazon Music, Apple Music, Spotify and YouTube Music.

To begin with, he didn’t need to be a musician. He uploaded hundreds of thousands of tracks created from scratch by artificial intelligences to these applications. They were produced and put online at a rate of several thousand tracks each week.

Michael Smith gave them names with an algorithm that automatically generated the titles of each of the pieces. “Title” is an overstatement since they were essentially made-up words with no particular meaning: Zymogenic, Zymoplastic, Zymotechnical Or Zymes ; nothing really artistic. Even the names of the “performers” were generated on the assembly line. Without any real search for credibility: they were called for example “Calm innovation”, “Calm Identity”, “Calorie Event” or “Calm Baseball”.

All that remained was to create batteries of user accounts, again in thousands of copies, to constitute the supposed audience of listeners. Michael Smith subcontracted this activity to foreign service providers who carry out large-scale repetitive tasks on demand. They developed the fake profiles from email addresses purchased in bulk.

Thanks to VPN technology, virtual private networks that allow you to modify the geographic origin of a connection, the platforms had the impression that the registered users came from multiple places across the planet. In order to go unnoticed. Then began the waltz of automatons: machines click on digitized pieces to create audience volumes, paid for by the royalties owed to composers and performers.

In addition to this type of fraud, digital music merchants are concerned about these compositions generated by algorithms: according to them, they weaken and relativize the reason for a subscription, especially when it is paid. And so they regularly carry out purges. Thus in the spring of 2023, Spotify removed around 7% of the tracks that had been published via Boomy, a start-up allowing you to create music on the fly. That’s the equivalent of tens of thousands of pieces.

In the emails that investigators found on his computers, the scammer explained in 2017 that his 20 machines at the time requested more than 630 songs daily. Moreover, as early as 2018, he had received warning messages from several platforms that regularly threatened to remove titles deemed suspicious. Which was done for some of them.

His initial idea was to consider that concentrating listening on a single title that would attract a billion views would be much more visible than spreading this same billion connections over tens of thousands of songs across the general catalog of the major platforms. He himself, in his messages with his suppliers, worried about the need to continuously produce “tons of songs”, in his words, in order to pass under the radar of the broadcasters’ surveillance.

While the means used here are virtual, the penalties incurred are very real. Michael Smith faces several charges, including wire fraud and money laundering. With penalties of up to 60 years in prison. For an estimated gain of 9 million euros.