Man Arrested After ‘Earning Millions’ From AI Music Tracks

A US man has been accused of fraudulently earning more than $10 million (£7.6m) in music-streaming royalties using hundreds of thousands of AI-generated tracks that he streamed to fake listeners.

Michael Smith, 52, of North Carolina, devised a “brazen” scheme to publish thousands of tracks per week on music-streaming services, playing them billions of times using fake bot accounts he programmed, according to an unsealed indictment.

After initially trying to produce large numbers of songs himself or with other artists, Smith teamed with the unnamed chief executive of an AI music company and a music promoter who supplied him with a huge volume of fake songs over a period of seven years, the indictment said.

“Smith stole millions in royalties that should have been paid to musicians, songwriters, and other rights holders whose songs were legitimately streamed,” said US attorney for the Southern District of New York Damian Williams.

‘Instant music’

Smith was arrested last week on charges including wire fraud, wire fraud conspiracy and money laundering conspiracy, with prosecutors saying it is the first case of the kind they have handled.

Prosecutors said that following their investigation, which involved the FBI, Smith was set to “face the music”.

The fake tracks would arrive to Smith with names such as n_7a2b2d74-1621-4385-895d-ble4af78d860.mp3, and he would assign them song and creator names such as “Zygopteris”, “Zygopteron”, “Zygopterous” or “Zygosporic”, prosecutors said.

Smith’s fake band names included “Callous Post”, “Calorie Screams” and “Calvinistic Dust”, with tracks named “Zygotic Washstands”, “Zymotechnical” and “Zygophyllum”.

“Keep in mind what we’re doing musically here,” the AI executive wrote in a 2019 email to Smith. “This is not ‘music,’ it’s ‘instant music’ ;).”

Smith created as many as 10,000 streaming accounts using email addresses he purchased online, eventually outsourcing the process to paid assistants.

He created software to stream his songs to different computers, making it appear that listeners were tuning in from various locations, prosecutors said.

Billions of streams

In an email to himself in 2017, the year prosecutors said he started the plan, Smith estimated he could bring in up to $1.2m a year in royalty payments.

The fake streaming activity was spread across a huge number of fake songs, with no one song streamed a large number of times, in an effort to avoid detection.

By June 2019 Smith’s scheme was bringing in about $110,000 per month, a portion of which went to his co-conspirators, prosecutors said.

In an email in February Smith said he had reached 4 billion streams and $12m in royalties since 2019.

The AI music-generation technology improved over time, making it harder for music streaming companies to spot, the indictment said.

In October 2018 a platform told Smith it had received “multiple reports of streaming abuse” and planned to remove his songs from its stores.

Copyright concerns

“This is absolutely wrong and crazy!” Smith replied, the indictment said. “There is absolutely no fraud going on whatsoever! How can I appeal this?”

AI has raised concerns in the entertainment industry as content-generation tools are based on training data that often includes copyrighted work, for which the original creators are not compensated.

Earlier this year well-known music artists signed an open letter calling for an end to the “predatory” use of AI in the music industry.

In June major music labels sued two AI start-ups over alleged copyright infringement, in the latest case that may provide a legal test for generative AI companies’ business models.