He made $10 million from AI music streaming and bots

Michael Smith, a 54-year-old musician from North Carolina, came up with an audacious plan that he successfully implemented for several years. The man established cooperation with the head of an AI music company and an anonymous music promoter, from whom he purchased wholesale quantities of generated songs. Then he uploaded them to popular streaming platforms and made fake plays. To bypass service providers’ anti-fraud systems and not arouse suspicion, the fraudster masked bot traffic using a VPN.

The scale of this practice was almost industrial. At its peak, Smith was using over 1,000 bot accounts spread across 52 cloud servers. His own calculations, sent in one email, showed that bots played songs over 660,000 times each day. At an average rate of half a cent per stream, this generated a daily profit of over $3,300.

In messages to his partners, Smith emphasized that the key to fooling the algorithms was having a gigantic number of songs, each of which generated only a small number of plays. Already in 2017, when Smith started his practice, there were artificial intelligence models and tools that could generate music. Although this technology did not yet allow for the creation of realistic vocals as freely as today, it successfully enabled mass production of, among others, instrumental pieces. Michael Smith used this mechanism.

In February 2024, the musician boasted in emails that his system had generated over 4 billion streams and $12 million in profit since 2019. However, the American prosecutor’s office calculated losses of USD 10 million and emphasized that although the songs and listeners were fake, the money withdrawn reduced the pool of funds for real artists.

Now the 54-year-old has pleaded guilty to wire fraud. The man agreed to forfeit property in the amount of over USD 8 million, and for committing a federal crime he faces a sentence of up to 5 years in prison. Initially, he faced up to 20 years in prison for each of the charges, including: money laundering.

The case of the American musician is just the tip of the iceberg. Streaming services are flooded with music generated by AI – not only simple, ambient instrumental lines, but increasingly “recordings” with human vocals, difficult to distinguish from live performers.

In November 2025, Spotify announced that it had removed from its library as many as 75 million songs classified as spam generated by AI algorithms. At the same time, the case of the Velvet Sundown project was widely reported, gaining over a million fans before the creators revealed that it was an artistic joke produced by algorithms.