AI music-streaming fraud defendant seeks probation despite two-year recommendation

Michael Smith, who pleaded guilty to a wire-fraud conspiracy involving artificial-intelligence music and bot-generated streams, is asking a federal judge to impose probation rather than prison. Federal officials have recommended a two-year sentence, while Smith has agreed to forfeit more than $8 million in proceeds. Smith’s lawyers argued that no individual artist or songwriter sustained “perceptible harm,” according to Billboard. They characterized the impact on any one artist as tiny and immeasurable, comparing a diverted stream’s effect on other artists’ shares of total plays to “a drop in an ocean.” The defence nonetheless acknowledged that Smith used increasingly deceptive methods and knew he was cheating the system. It said he had accepted full responsibility and was profoundly sorry for his crimes. The offence carries a maximum prison term of five years. Federal sentencing guidelines put his range at 46 to 57 months, but the Presentence Investigation Report recommended 24 months. He was arrested in 2024 after prosecutors accused him of running thousands of bots that continuously played music on Spotify, Amazon Music, Apple Music and YouTube Music from 2017 onward. The government said Smith initially used human-created recordings before switching to mass-produced AI tracks, generating thousands of songs and billions of plays. The indictment alleged that an unnamed AI-music-company chief supplied songs in return for a share of the proceeds. Hundreds of works registered to Smith list Alex Mitchell, the founder of Boomy, as a co-writer, although Mitchell has not been charged. U.S. Attorney Jay Clayton rejected the premise that the harm was merely theoretical, saying that while “the songs and listeners were fake,” the millions of dollars taken were real. The case highlights a central feature of streaming fraud: royalty payments are drawn from a finite pool intended for creators, artists, labels and publishers. The World Intellectual Property Organization has said artificial streams divert money from that pool, while manipulated play counts can also skew recommendation systems and consumer data, making legitimate artists harder to discover. AI can make such schemes harder to detect by allowing fraudsters to upload huge numbers of tracks and spread plays across them, rather than concentrating activity on a small catalogue. WIPO described Smith’s case as an alleged extraction of more than $10 million in royalties through AI-generated songs and bots, a figure distinct from the more than $8 million he agreed to forfeit. The scale of the wider problem remains disputed. A 2021 study by France’s National Music Center estimated that 1% to 3% of all streams were fraudulent, while fraud-detection company Beatdapp estimated about 10%. Deezer caps royalty-generating plays from a single user at 1,000 streams, with a sharply lower payment rate after that threshold. Smith’s case, United States v. Smith, is before Judge John G. Koeltl in the U.S. District Court for the Southern District of New York.