Neural networks helped the musician cheat to earn $10 million on streaming services

Prosecutors say the North Carolina resident used artificial intelligence to create hundreds of thousands of fake songs from fake bands, then uploaded them to streaming services where they were enjoyed by an audience of fake listeners. Now the musician faces at least 20 years in prison.

Image source: Alexander Sinn/Unsplash

According to The New York Times, 52-year-old musician Michael Smith was accused of fraud related to the manipulation of streaming services. Prosecutors say he used neural networks to create hundreds of thousands of fake songs, which he then posted on popular platforms such as Spotify, Apple Music and Amazon Music. As a result of his fraudulent scheme, Smith made at least $10 million by faking royalties and deceiving listeners who did not actually exist.

The fake artist scam resulted in serious charges including email fraud and money laundering conspiracy. Prosecutors said Smith used software to automatically stream his generated songs, creating the illusion that there were real artists behind them. He allegedly had fictitious bands such as “Callous Post”, “Calorie Screams” and “Calvinistic Dust”, which released tracks with unusual titles like “Zygotic Washstands” and “Zymotechnical”.

Prosecutors emphasized that “Smith stole millions in royalties that should have been paid to musicians, composers and other assignees whose songs were legitimately aired.” As a result, this case became the first criminal case related to manipulation of music streaming brought by the U.S. Attorney for the Southern District of New York. If Smith is found guilty, he could receive up to 20 years in prison on each charge.

Smith’s scheme was carefully thought out. He created thousands of fake streaming accounts by purchasing email addresses on online platforms. Having up to 10,000 such accounts, due to the labor-intensive nature of the process, he attracted other users (participants) to provide paid assistance in their creation. Smith also developed software to play his songs over and over again from different computers, making it appear as if real listeners were following the music. In 2017, according to prosecutors, he calculated that he could stream his tracks 661,440 times a day, which provided him with an income of more than $3,000 a day.

Smith started out by putting his own original music on streaming platforms, but soon realized that the number of his tracks was not enough to earn significant royalties. Attempts to use music owned by others and offer services to promote such compositions have not been successful. In 2018, he teamed up with the head of an AI music company and a music promoter to create a huge catalog of fake songs that he uploaded to streaming platforms.

As of June 2019, Smith was earning about $110,000 a month, some of which went to his co-conspirators. In one of his letters in February of this year, he boasted that he had reached 4 billion streams and $12 million in royalties since 2019. However, when streaming companies began to suspect him of cheating and notified him that they had received “several reports of abuse,” Smith responded by saying, “This is absolutely not true, this is crazy! There is no scam going on here! How can I appeal this?