Michael Smith of Cornelius, North Carolina, was sentenced to 18 months in prison for using bots to collect about $8 million in royalties from AI-generated music. He must forfeit $8,091,843.64 earned between 2017 and 2024 and will face two years of supervised release. The case matters because it draws a line between legal creation of AI tracks and illegal monetization through fake streams.
How the bot-driven royalty scheme worked
Smith uploaded hundreds of thousands of AI-generated songs to Spotify, Apple Music and Amazon Music, then inflated listening statistics to billions of plays. Platforms paid royalties under per-stream plans, which turned artificial listening into real revenue. He was charged in September 2024 with wire fraud conspiracy, wire fraud and money laundering conspiracy. In March 2026 he pleaded guilty to a single count of conspiring to commit wire fraud, avoiding the maximum five-year term.
According to the account cited in the report, Smith controlled 52 cloud service accounts with 20 bots each, giving him 1,040 agents connected through VPNs. Each bot could stream roughly 636 songs per day, producing about 661,440 daily plays across the network. The US Justice Department estimated the scale even higher, saying Smith deployed up to 10,000 bots at a time. In April 2023 his music logged 80.9 million plays on YouTube Music, compared with 9.3 million for Taylor Swift in the same period.
The prosecution did not target AI music itself, which remains lawful to create and distribute on streaming services. Liability arose from misrepresentation of popularity and diversion of royalty pools from legitimately streamed musicians to automated listening. The Justice Department described streaming fraud as moving funds away from songwriters whose tracks were played by real consumers. That distinction explains why the sentence focused on fraud mechanics rather than generative technology.
What the verdict means for AI content business
For companies that publish or license AI audio, the ruling clarifies that distribution scale alone creates no income without genuine demand. Apple Music Vice President Oliver Schusser said more than a third of tracks on the platform are 100 percent AI, yet they account for below 0.5 percent of usage. Small catalog owners therefore gain little from mass uploads, while large rights holders keep most payouts tied to real listening. Investment shifts toward promotion, playlist placement and audience verification.
Detection and proof of origin are becoming part of procurement and compliance checks for media platforms. Google said it has watermarked 240,000 years worth of AI-generated audio and 180 billion AI images and videos with its SynthID Detector tool. Initially for pro users, the detector is now available to anyone and works with content from Google, OpenAI, Nvidia and Kakao, with Apple to follow. Buyers should still ask vendors how synthetic content is labeled, how bot traffic is filtered and how royalty reports are audited, since watermarking does not by itself stop fraud.
The marker to watch is whether streaming services publish new fraud filters, disclosure rules or penalties after this sentence. A second signal is wider adoption of detectors such as SynthID across distributors and rights societies. If fake-play volumes fall and royalty disputes decline, the Smith case will have set an operational precedent for AI catalogs.
