The Recorded Music Industry Wants Labels on AI Music. How Could That Work? (FAQ)

Skip to main content The record industry has proposed labels for fully and partially AI-generated tracks on streaming services, but the implementation could take a while. AI Billboard On July 10, key stakeholders in the recorded music industry came together to announce a new system for identifying fully and partially AI-generated songs. Since the release of Ghostwriter’s so-called “Fake Drake” song “Heart On My Sleeve” in May 2023, the industry has been grappling with how to deal with AI-generated material on streaming services, with DSPs each proposing their own set of guidelines. That includes anything from tagging AI-generated songs as such, to enhanced streaming fraud detection, to demonetization. But Sam Duboff, global head of policy, music business at Spotify, told Billboard in 2025 that “it’s still early days for AI tech” — leaving many platforms hesitant to create any stringent long-term rules that could lead to unintended consequences. Related Though the technology might be relatively new, French streaming service Deezer says the tide of fully AI-generated songs delivered daily to its platform is rising fast. As of Tuesday (July 21), Deezer claims up to 50% of all songs delivered daily are fully AI-generated. Pair that with a recent study from Luminate that noted 42% of respondents are less interested in songs when they learn generative AI was involved in their creation, and it seems to many like the right time to take another step forward. Still, streaming services like Tidal and Apple Music have expressed the need for labels and distributors, who are the ones delivering AI content to streaming platforms, to step up and help. “I really need the content providers and the labels to take responsibility,” Apple Music chief Oliver Schusser previously told Billboard. A Tidal spokesperson echoed that sentiment in June, stating, “The responsibility to identify and tag AI-generated content should not rest with Tidal alone.” Now, the recorded music industry is heeding the call with its own ideas, creating a system that could be uniformly implemented across platforms — that is, if all the streaming services and related stakeholders sign on. The road ahead, however, may still take months to figure out, and three sources have told Billboard they have concerns that the AI-assisted tag proposal, in particular, may be a step too far as AI becomes a more normal part of the creative process. So, how will the process work? How will songs be tagged correctly? And how does something like this compare to previous instances in music history, such as the parental advisory sticker, that have sought to provide warnings to listeners about the content they’ll be hearing? Here are the answers to some of those questions. Who Is Behind This Initiative? This announcement was put together by the Recorded Industry Association of America (RIAA), the American Association of Independent Music (A2IM), the Worldwide Independent Network (WIN), the European Independent Music Companies Association (IMPALA), the Recording Academy, SAG-AFTRA and the Human Artistry Campaign. How Much AI Music Is There, Really? It’s impossible to say how much AI music exists today, but information from Suno, Apple Music and Deezer can help illustrate the current scale. According to an investor pitch deck for Suno in late 2025, the AI music giant generated 7 million songs daily — or an entire Spotify catalog’s worth of music every two weeks. Many of these AI songs created on Suno, however, do not leave the platform, and each time a user prompts Suno for music, the model replies with two song options. In a previous interview with Billboard, Apple Music’s Schusser said that “more than a third” of their monthly intake was “100% AI” generated sound recordings, and that the company is now using an in-house detection tool to track AI use internally. French streaming service Deezer claims to see a higher count of AI tracks. Since January 2025, the company has been the most transparent and public about its findings regarding AI music on its platform. Back then, the company first reported that 10% of daily uploads were fully AI-generated recordings, according to its AI detection tool. As of Tuesday (July 21), however, Deezer now reports that the figure has topped “over 50%” of daily uploads on heavy days. Related Deezer and Apple’s tallies still might fall short of the true impact of AI music on platforms today. Both only count 100% AI-generated recordings, meaning they do not count anything that is partially human-made. Manuel Moussallam, director of research for Deezer, previously told Billboard in June 2025 that the company’s findings regarding the amount of AI music on its platform have climbed so high in part because of the further adoption of AI music tools, also adding that the company’s data “got better.” Moussallam further explained, “Part of the stuff we were catching as AI-generated in January [2025], we weren’t totally sure about, so we were very conservative in the numbers we reported. We didn’t want any false positives.” Haven’t Streaming Services Already Introduced AI Labels on Their Platforms? Yes, some of the streaming services have already created their own policies for tagging or labeling AI-generated content. Tidal’s system is the most similar to what has been proposed by the recorded music industry. Announced in June, Tidal says it’s tagging AI songs based on a mix of distributors and labels identifying AI-generated music before uploading it to the platform and using AI detectors. At first, it will only tag songs that are identified as 100% AI-generated, but as “AI detection methods become more reliable,” it will “expand” the tag to music “that is substantially AI-generated,” the company said. Deezer also shows visible tags for fully AI-generated songs. The difference between Deezer’s plan and what the industry proposes is that Deezer only relies on a proprietary AI detection system to automate tagging, rather than asking artists and labels to voluntarily disclose AI use. It also does not have a tag for partially AI-generated songs, which is part of the industry coalition’s proposition. As of April, Spotify also has an AI crediting system. This policy does not create a visible label on AI songs, but if a user views a song’s credits, AI can be listed for any part of the song’s creation, whether that be for just the drums, just the vocals or everything — just as people involved in a song can be listed. Spotify’s system depends on artist disclosure, and a blog post announcing the policy noted that “the absence of a credit doesn’t mean AI wasn’t used…we know this isn’t a complete solution on its own.” Apple Music also has an AI credits system called “Transparency Tags,” launched in March. The platform calls this a “delivery requirement” for labels and distributors and allows them to credit AI use in the following categories: sound recording, composition, artwork and music video. Related Who Would Be Responsible for Adding These Labels? The coalition’s proposal says that this is “voluntary track labeling,” meaning it is up to the music creators to disclose whether they’ve created a song fully or partially with the help of AI. There is no mention of the use of AI detection, and one source close to the matter tells Billboard there are still concerns about false positives from current AI detection tools on the market. False positives could lead to major headaches, not just for streaming services but for artists who inadvertently and incorrectly were flagged for AI use, and they would face an uphill battle trying to prove otherwise given the volume of content on any given DSP. Is This System Definitely Getting Implemented? It is unclear if streaming services will be adopting this system, as they did not sign on to the announcement with the recorded music industry. Soon after the announcement was made, however, the Digital Media Association (DIMA), which acts as a trade organization for music streaming services, signaled some support for it, saying, “We are following today’s announcement closely and look forward to receiving more detailed and accurate AI metadata, which will strengthen our ability to give fans the transparency they deserve… Our members look forward to continuing with the labels, producers, artists and distributors, as well as other industry stakeholders and standards bodies such as DDEX, to build a robust supply chain in which consumers can trust.” A spokesperson for Deezer also signaled support for the announcement, saying, “Deezer is ready to support the development of an industry-wide framework. This includes key considerations around the use of training data for AI models, ensuring that all rights holders are fairly remunerated. We’re looking forward to continuing our collaboration with the wider music ecosystem to create fair and practical standards for AI in music.” If the recorded music industry, the artists they represent and the streaming services can all align on the record industry’s proposal, it will likely still take time to see these tags appear online, as each platform has its own specifications that would need to be taken into account. It’s also unclear how this system would work with the other policies streaming services already have for AI music. Would the previous policies be tossed out to implement this standardized one? Would both policies co-exist? Only time will tell. How Will This System Differentiate AI-Assisted Songs and Fully AI-Generated Songs? In theory, there would be two separate tags: one uppercase “AI” icon for “AI-generated” tracks, and a lowercase “ai” icon for “AI-assisted” tracks. The uppercase label is for content fitting the following criteria: “lead vocal performance generated by AI, key instrumental performance generated by AI, or entirely prompt-generated AI music,” according to the announcement. The lower case label is for content that uses “generative AI… for some expressive elements. Humans performed the lead vocal and primary instruments.” The distinction between what constitutes AI-generated and AI-assisted works remains hotly disputed in the music business today. As former record executive and current AI researcher Drew Thurlow pointed out in a guest column for Billboard last week, the term “AI-assisted” is “almost impossible to define” — and with the number of producers, writers, engineers and mixers involved in the modern recording process, almost any song commercially released these days could have an AI element in it, whether the artist and label are aware of it or not. The system, then, would require complete transparency throughout the entire writing and recording chain, a process that would be Herculean and probably not achievable in reality. The coalition’s proposal doesn’t directly address this issue, meaning it would likely be the subject of debate if it were to be implemented in any way. How Is This Initiative the Same, or Different, From the ‘Parental Advisory’ Labeling System That Was Introduced in the 1980s? The “Parental Advisory: Explicit Content/Lyrics” warning label was introduced in 1985 after being championed by a group of parents and teachers who were concerned by many forms of inappropriate content in songs of the day. The idea was hotly contested. Artists like John Denver, Frank Zappa and Dee Snyder argued against the proposal at a Senate hearing, while Mike Love of the Beach Boys provided financial support to assist with its adoption. In the end, the Parental Advisory tags won out, and through a deal with the RIAA, they were added to albums starting in November 1985. Both the AI tag proposal of today and the Parental Advisory movement of 40 years ago would cause some songs to list warning labels, providing transparency for music listeners before even pressing play. Plus, both initiatives contain some level of support from the RIAA. The difference is that the AI tagging proposal does not appear to be as contested as the explicit content tags of yesteryear; for example, big-name artists have yet to speak out against it. Another difference? Parental Advisory labels were devised to warn parents about inappropriate themes and lyrics in songs, whereas AI-generated songs don’t pose any threat to younger listeners. Sure, it may be off-putting that a person used AI to create a song, but is AI use something that could harm consumers in the same way sexual or violent lyrics could? Arguably no. Another consideration: The adoption of explicit content labels was essentially a requirement from retailers in exchange for carrying those titles, meaning that labels and artists had a commercial incentive to conform. Simply, with no explicit content labels, they couldn’t get their albums into stores. There’s no such incentive for artists and labels to do so with AI-generated or AI-assisted content. Instead, it would simply be on the honor system — which could become ripe for exploitation. Daily newsletters straight to your inbox Sign Up More From Pro