The creators of AI-generated music that manipulates streaming services to deprive artists of revenue are nothing but a bunch of thieves, says the International Federation of the Phonographic Industry (IFPI). The music industry trade body is issuing stronger guidelines to distributors to prevent fraudulent AI music from reaching streaming services.
“Streaming fraud misleads fans and deprives legitimate artists and songwriters of income that should rightfully flow to them. This is theft, plain and simple,” says Victoria Oakley, CEO of IFPI.
The IFPI Streaming Integrity Initiative involves more than two dozen labels and distributors. It includes industry powerhouses like Sony, Universal and Warner. Under the new guidelines, distributors are expected to verify rights ownership, check content for copyright infringement and fraudulent activity, assess AI-related risks and take action against repeat offenders.
Monitoring for AI content also is likely to be done on a more technical basis. “By integrating deep, multi-signal fraud detection into our music OS, Absolute is enabling rights holders to identify potential irregularities early,” said Simon Wills, managing director of Absolute Label Services.
Spotify says it will review suspect profiles whose names and images appear to be AI-generated identities.
The IFPI move comes as AI makes it easier to create large volumes of music and to use automated accounts to artificially inflate stream numbers. Industry estimates suggest nearly 10% of music streams are fraudulent. Some streaming services report higher numbers. French streamer Deezer, for example, said fully AI-generated tracks accounted for more than half of daily uploads at their peak in June, while a large share of streams involving those tracks showed signs of fraud.
Streaming fraudsters create a large number of music tracks and then use bots to play them repeatedly. AI has made the creation of music tracks much easier, as they can be generated using simple prompts. Royalties are distributed from a pool based in part on the number of streams. Fake plays reduce the share available to legitimate artists.
The amount of money in play can be significant. In one criminal case in North Carolina, a fraudster created hundreds of thousands of AI-generated songs and thousands of accounts on streaming services. The scheme netted the now-convicted fraudster $8 million in royalties.
Record labels are also going after alleged AI fraud in the courts. Universal Music Group (UMG) has brought suit against DistroKid, which presents itself as a music distributor of artist-created work across several streaming platforms. UMG alleges DistroKid built an “AI-slop pipeline” that continued even after DistroKid was informed of alleged infringement involving artists’ work. The UMG suit cites the example of a “Lofi Chill” artist that produced 4,562 tracks within a 12-month period.
“The consequences of this deception fall on legitimate players in the music industry, including artists, consumers, digital services and plaintiffs,” said UMG in its suit, noting that DistroKid bore none of the costs associated with producing music by authentic artists even as it allegedly presented itself in that light. “A business built on such deception makes competition unfair,” concludes UMG.
Some streaming services also are taking action by introducing labels for AI-made tracks, a move that appears to be endorsed across the music industry. Music streamer Tidal allows AI-generated music on its service but says it does not knowingly attribute royalties to wholly AI-generated music.
Meanwhile, some observers think that improvisational, algorithmic-resistant genres like jazz may see an upswing in popularity. Additionally, a Luminate survey found that younger Gen Z and Gen Alpha music fans are increasingly negative toward AI-generated tunes.
The music industry effort to contain AI-generated music is bottom-line focused, as might be expected. But perhaps far more concerning is the effect AI-generated tunes are having on people. A Deezer survey of 9,000 people, conducted across eight countries, found that 97% of people couldn’t tell the difference between AI-generated and human-made music in a blind listening test. The core challenge: If 97% of people can’t tell the difference, how do you convince them the difference matters?
