We label almost everything. We label food as organic, artificially flavored, genetically modified or sugar-free. And we label advertisements so consumers know when someone has been paid to promote a product. Movies come with ratings. Additionally, social media platforms increasingly identify altered photographs and synthetic videos.
Those labels do not necessarily tell us whether something is good or bad. They simply tell us what it is. Music deserves the same basic transparency. If a song was principally written, performed and produced by artificial intelligence, the listener should know that. A radio programmer, playlist curator or music fan should not have to conduct a forensic investigation to determine whether the singer they are hearing is an actual person.
Establishing Clear Categories
That part seems easy. However, the more difficult question is where the label begins and where it ends. The music industry recently proposed two track-level designations: “AI-Generated” for recordings in which artificial intelligence created all or most of the important creative elements, and “AI-Assisted” for music that remains substantially human-created but includes meaningful generative-AI contributions. The proposal has support from organizations representing major labels, independent labels, artists and performers.
It is an important first step because “made with AI” has become an overly broad description. Specifically, using artificial intelligence to remove background noise is not the same as using it to invent a singer. Automated mastering differs significantly from entering a prompt and having a machine write, perform and produce an entire song. Using technology to clean up a human vocal should not put a recording in the same category as generating the voice, personality and performance from scratch.
The clearest test may be this: Did AI help a human make the recording, or did AI make the recording for the human?
Navigating the Gray Areas
There will still be gray areas. A human might write the lyrics while AI creates the melody. A producer might record a real vocalist but generate the backing musicians. Furthermore, someone might create an AI track and then use human engineers to edit, mix and master it.
Recent research argues that a simple “AI or human” classification no longer reflects how music is being produced. Notably, AI can now participate at different stages, including composition, lyrics, vocals, arrangement and mastering. That means the eventual labeling system will probably need to disclose where AI entered the process, not merely whether it appeared somewhere.
However, the existence of gray areas should not become an excuse to ignore the obvious cases. Fully AI-generated music needs to be labeled. There is no legitimate reason to conceal it.
Addressing Scale and Transparency
The scale of the issue makes transparency even more important. Deezer reported that AI-generated tracks exceeded 50% of its new uploads at peak levels in June, totaling approximately 90,000 tracks per day. The concern is not simply whether an individual AI song is good enough to attract an audience. Rather, it is whether automated production can overwhelm the music ecosystem through sheer volume.
Imagine opening your station’s new-music folder and discovering that half the submissions were created by machines. Some might be entertaining. Some might even be hits. But programmers should know whether they are supporting an actual artist who can perform, interview, tour and build a relationship with the audience.
Another recent study found that 92.7% of AI-generated tracks received fewer than 1,000 Spotify plays during the first five months of 2026. In comparison, 67.5% of human-created tracks achieved the same threshold. Researchers described a “spray and pray” strategy in which AI operators release enormous quantities of music across multiple genres in the hope that something gains traction.
This presents another problem: responsible artists and labels may willingly disclose their use of AI. Meanwhile, operators uploading thousands of anonymous tracks may be the least likely to tell the truth. Eventually, disclosure will need to be supported by standardized metadata, credible detection systems and consequences for deliberately mislabeling a recording.
Finding Balance
Still, the label should not become a scarlet letter. AI will be part of legitimate music production. Artists will experiment with it, producers will use it and the technology will continue to improve. “AI-Assisted” should not automatically mean fake any more than “recorded digitally” means nobody played an instrument.
A label provides information. It should not pass artistic judgment. Importantly, radio also has a legitimate stake in this discussion. Programmers should know whether an artist and performance are real, whether a voice was digitally replicated and whether the named artist consented to the recording. A station would not necessarily need to announce the label every time the song plays. But that information should be available before radio decides to add the record, interview the supposed artist or invest its credibility in building the act.
The Listener’s Right to Know
Perhaps listeners will embrace fully synthetic performers. Perhaps they will reject them. Ultimately, many people may care more about whether they enjoy the song than how it was produced. That decision belongs to the listener. However, listeners cannot make an informed choice when the process is intentionally hidden from them.
The music industry does not have to fear artificial intelligence. It does, however, have to label it honestly.
Barrett Media produces daily content on the music, news, and sports media industries. Sign up for our newsletters to stay updated and get the latest information right in your inbox.
David Hill serves as a Music Radio Editor, Columnist and Features writer for Barrett Media. A radio lifer with more than 30 years behind the mic, in the control room, and in the program director’s chair, David’s career spans influential stops at brands such as WIYY 98 Rock, WBAL-AM, and 99X. He has worked across multiple formats and ownership groups, including iHeartMedia and Cumulus Media, developing talent, breaking music, and navigating every major industry shift from diary to PPM and terrestrial dominance to streaming disruption. When he’s not writing or analyzing the industry, Dave runs The Tune Farm, a marketing firm built to help artists and brands grow audience the same way great radio always has—by creating connection, not just impressions. He can be reached at David@BarrettMedia.com.