Music Streaming Fraud: What the US Copyright Office Is Asking About AI Music and Royalties
Short answer: the US Copyright Office is asking the music industry how streaming fraud works, who is placed to stop it, and how it relates to AI-generated music. Comments are now due by December 18, 2026. If you release music made with AI tools, the inquiry hints at what distributors and platforms may ask you to show: who made the track, how its plays were earned, and what records back that up.
What is the US Copyright Office asking about streaming fraud?
The Office published a notice of inquiry on October 7, 2026 (Docket No. 2026-6), after Representative Scott Fitzgerald asked it to "examine the prevalence of streaming fraud across digital platforms and how it may be affecting the music industry as a whole." The notice lists ten subjects, from how fraud is carried out to possible voluntary, statutory or regulatory fixes.
It defines streaming fraud as using "deceptive, inappropriate, or unpermitted methods to artificially increase the number of streams a song or album receives." The methods it names include bot farms, click farms, playlist stuffing and account hijacking. It also describes "copyright fraud", such as uploading manipulated versions of other people's works.
Comments were first due on November 23. On November 5 the Office extended the deadline: initial comments are now due by 5:00 p.m. Eastern Time on December 18, 2026, and reply comments by January 29, 2027. Comments are filed on regulations.gov and posted publicly.
Where does AI-generated music come in?
Question 4 asks directly what data exists on the relationship between streaming fraud and "the effects of AI-generated [music]" on licensing and royalties. The notice does not say AI music is fraudulent. It asks for evidence.
The reason it matters is the payout model the notice describes. Under a pro-rata system, the more a song is played, the more its rightsholders receive, so the Office calls "the dilution of the royalty revenue pool" the primary harm. Whether AI-generated catalogues change that picture is exactly the data Question 4 asks commenters to supply.
Our reading: the inquiry treats AI as one factor to measure, not as the definition of fraud. A track made with AI and streamed by real listeners is a different case from any track pushed by bots.
What happened when a royalty society tried to check for AI?
South Korea offers a recent example. On October 22, the Korea Music Copyright Association (KOMCA) said it was withholding about 24 billion won ($18.1 million) in royalties claimed by four rights holders for 802 songs, over allegations that tracks were inserted into YouTube Shorts to inflate usage. KOMCA also sent 791 related tracks it had obtained to overseas AI music analysis firms. According to The Herald Business, the findings varied across providers, and KOMCA said it could not determine AI authorship from those results alone.
These are allegations under review. Police declined to forward the case to prosecutors in June, citing insufficient evidence, and KOMCA has asked for a review of that decision. The four rights holders' positions could not be confirmed.
Our reading: when detector scores disagree, the decision moves to other evidence, such as how the plays were generated and how the songs were made. We covered how to read a single score in AI music detector false positives.
Which questions in the inquiry affect independent artists most?
Inquiry question (summarised) | What it could mean for your release | What to keep on file |
|---|---|---|
Q4: streaming fraud and AI-generated music | More questions about how a track was made | Tools and plans used, project files, the parts you wrote or played |
Q6b: trusted distributors, content and customer verification, royalty clawback, fines, removal | Identity checks, held payouts or takedowns if plays look unusual | Distributor account details, promotion invoices, where your listeners came from |
Q7c: metadata standardization | Credits and AI disclosure fields may carry more weight | Consistent credits and any AI declaration you submitted |
Q7b: a shared database of known fraud | A flag could follow an account across services | Records showing how each promotion was bought and run |
This table is our own reading of the notice as of November 13, 2026. The Office has not proposed any rule, and the right-hand columns are suggestions, not requirements.
The notice also reports that some scammers keep going by "bouncing from distributor to distributor," and that at least one distributor has proposed a worldwide database of people found abusing the system. For how distributors already ask about AI, see AI music distribution and AI credits.
What can't this guide tell you?
A notice of inquiry gathers information. It does not create obligations, and nothing in it tells you whether a specific release is at risk. What Congress or the Office does with the comments is not yet known. The KOMCA case is still under review, and we make no finding about the songs or the people involved.
FAQ
Is AI-generated music considered streaming fraud?
Not on its own. The notice defines fraud by how streams are inflated, and it quotes the congressional request that "there is nothing inherently wrong with using AI as a creative tool in music development." It asks for data on how AI music relates to fraud.
Can independent artists submit comments?
The notice solicits information from the public, and comments are filed on regulations.gov. Initial comments are due December 18, 2026.
Can an AI detector prove a song is fraudulent?
A detector estimates whether audio looks AI-generated. It says nothing about how plays were earned, and in the KOMCA case different providers disagreed.
Want a reference reading to keep on file for a track made with AI tools? See how it reads on MusicScanner's AI analysis index (open beta).
It is for reference only, not a fraud or rights decision.