Xania Monet released the single “How Was I Supposed to Know?” last year, and it immediately went on its way to the Billboard charts, reaching number 30 on Adult R&B Airplay and number 20 on Hot R&B Songs. Another single “Let Go, Let God” peaked at number 3 on Billboard’s Hot Gospel Songs. With this momentum under her belt, she pumped out at least 20 more singles on top of four studio albums, and underwent a bidding war between various record labels, eventually signing with Hallwood Media for $3 million.
The catch? Monet is entirely made up of artificial intelligence (AI) — “she” isn’t real at all.
With the rise of AI usage across industries, the music sphere has inevitably been affected by unprecedented technologies. What started as individuals creating fake songs as internet memes has turned into users utilizing platforms like Suno AI and Udio to use generative AI to produce beats, vocals and even full songs with merely a text prompt — just like Monet’s generator, Telisha “Nikki” Jones, did. This has made it difficult for labels, curators and listeners alike to distinguish between methods of production for personal and copyright purposes.
While streaming platforms such as Spotify and Apple Music have introduced measures to transparently label music made with AI usage, AI songs, artists and overall usage are inconsistently disclosed. Consumers often value having a transparent understanding of the media they consume and being informed as to who or what they are supporting.
While there is no single foolproof way to identify AI music, as a listener, there are a few methods you can use together to critically examine the likelihood that a particular song or artist profile is AI to stay informed.
Let’s support real art made by humans, not art-stealing softwares.
On streaming services, keep an eye out for…
Unrealistic Productivity
When looking at an artist’s profile on any streaming platform, you can construct a timeline of what an artist has been up to release-wise. While a couple singles being released within a few weeks or even days of each other may not be indicative of anything, singles consistently being released in short times spans suggest production via AI. Similarly, if albums or EPs appear to be being put out back to back with minimal time in between, the artist may warrant further investigation, as that amount of productivity may be unrealistic for the average singer-songwriter. If you think about your favorite large artists, you may spend up to years waiting for their next album; handcrafted bodies of work take time.
AI Cover Art
Cover art for singles, EPs and albums are key artistic reflections of an artist’s work. As musicians spend hours of time and energy poured into their sonic work, cover artists and photographers also work hard to visually reproduce the artistic vision of the musical piece. With AI-generated illustrations increasing in abundance, they can also frequently be found accompanying other AI generations, including music. No matter how specific the prompt, these images consistently lack the human experience needed to visually adapt a music artist’s sonic vision, and discount the training, work, and passion cover artists and photographers pour into their craft.
While music accompanied by AI cover art on streaming platforms does not inherently mean the audial work is not original, if AI artwork appears consistently on artists’ profiles that have piqued concern for other reasons, it may serve as another indicator in that direction. AI artwork can typically be identified by out of place or inconsistent details, inconsistent lighting, and visual distortions or artifacts. It may also include broad, unintentionally generic imagery or hyperspecific, unrealistic photography.
The popular English rock duo Tears for Fears underwent scrutiny when the album art for Songs for a Nervous Planet was revealed to be made with generative AI.
Lack of Credits
Streaming platforms such as Spotify and Apply Music allow you to view individual credits on albums and songs. Individuals who worked on the tracks are credited in these descriptions, including vocalists, writers, producers, sound engineers and publishers. This both provides proper, fair attribution to specialists who contributed to the piece and forms a transparent public record of those involved in every track. Independent artists may (and probably do) have a smaller list of credits, especially those who sing, write, produce and publish their own music, but it is still uncommon to see a full-length discography without a single other collaborator, engineer, or copyright holder credited — especially when we take other AI-coated red flags into consideration.
Alone, missing credits may not indicate anything meaningful. Metadata can get lost, full information may not have been provided to Spotify, or credit information may be stored elsewhere. However, since credits for individuals and teams who work on music are a well-established component of released music, its readily-available absence may be a clue in the direction of AI generation.
On social media, look at…
Artist’s Online Presence
A physical presence via local shows or recorded interviews with magazines often easily reveal the face behind the music in your earbuds. However, smaller artists often utilize the internet in some way to market their music.
A well-run, consistently maintained social media status is often the standard for all artists, as it is one of the most effective ways to expand their reach to music listeners and acquire attention from others in the industry. Artists often post behind the scenes clips, livestream on platforms, or document their work and process to build a strong rapport with their audience.
Of course, some artists love their craft but equally value their privacy; a lack of complex online marketing or content creation may be the case, but is often accompanied by some music oriented posting across platforms or evidence of fan interaction online. A lack of original content altogether can be a useful signal when evaluating an artist’s authenticity.
Promotional Materials
A clean, polished online presence is not inherently unusual or a cause for suspicion; many artists brand themselves this way, and carry this atmosphere with the aid of a talented team of professionals behind the scenes.
Dissecting an existing online presence involves examining the artist’s specificity and consistency. Promotional materials that are unrealistically “perfect,” such as overly edited images or flawless videos with inconsistent details, may be an indication of generative AI composing the artist’s profile. This is exemplified in Xania Monet’s social media presence, where her body language and mouth movements are unrealistic and uncanny and all of her content is impossibly flawless, beyond even the swipe of Facetune.
Similarly, if all written materials involving the artist, such as their blog or any published interview, only include broad statements, lack anecdotes or offer limited information on collaborators or production, the line begins to blur and warrants further analysis.
With your saved music…
Deezer
Being aware of the music you know and love is equally as important as the new music you discover. As such, it may be worth analyzing the music already in your saved libraries. Recently, Deezer released a new software to detect AI songs within your existing playlists on streaming platforms. For those of us with large personal playlists or who frequently listen to platform specific algorithmically composed playlists, Deezer can play a crucial role in staying up to date on music we discover and immediately save on platforms themselves. Revisiting your playlists with Deezer can be a great step in the direction of becoming a more informed listener while platforms try to catch up with AI disclosure.
Online AI Music Detectors
If you are still feeling wary after examining a song or artist using the aforementioned methods, it may be worth running the music through an online AI music detection tool for an additional perspective. Readily available detectors such as SubmitHub and Beats To Rap On’s AI Music Detector analyze various components of a provided recording to provide an estimated probability that the piece may be AI generated.
As with any online detection tool, AI music detectors should be utilized with caution, as the methods each tool uses are proprietary and often unverified. False positives and negatives can be a risk, especially at intermediate results. For additional information, some detectors offer transparent methodological explanations, such as Beats To Rap On.
Extreme results from multiple detectors — consistently very high or low likelihoods of AI-generated components — may be more accurate, and help ground your previous assessments. As with all of these methods, detectors should not be used as a sole determinant, but as an additional piece of evidence in your investigation of transparency.
As AI-generated music projects become increasingly common, one thing is clear: people don’t want to support art that isn’t human made. Questions and concerns surrounding ethics, environmental impact and laziness often arise when discussing AI-generated art forms.
Just as many individuals take into account an artist’s personality, creative process and inspiration, and personal background of the artists they listen to, many desire to know if the music they are streaming was created by a human musician or is the product of an AI-generated project.
In today’s world, music listeners must reckon with an unprecedented challenge: identifying who or what is the true face behind the music in their earbuds. Fans have a right to have transparent information on the artists they choose to support.
While no single clue is sufficient to draw a conclusion, piecing together information based on productivity, credits, online presence and materials, and external sources can paint a better picture. Collectively, these details can help you determine if your recently discovered independent artist is a musician with a developing career, or an AI project generated from the click of someone’s keyboard.
