AI-generated music is facing new restrictions across streaming services and official charts as the recording industry begins setting rules for how artificial intelligence is disclosed and whether AI-made songs can compete for chart positions.
Australia’s ARIA Charts began excluding wholly AI-generated tracks from its Aug. 31 ranking, while the International Federation of the Phonographic Industry, or IFPI, has introduced similar eligibility standards across charts it manages in Latin America, the Middle East, Africa and Southeast Asia.
Streaming platforms are moving in the same direction. Apple Music plans to display “Made With AI” labels later this year, while Spotify will begin showing AI Persona badges on some artist profiles in mid-September.
The moves mark a shift from the industry’s earlier focus on copyright lawsuits to rules governing how AI music is identified, distributed and counted.
In July, IFPI, the Recording Industry Association of America, the American Association of Independent Music, SAG-AFTRA and other music industry groups announced a voluntary labeling framework that divides recordings into two categories: “AI-Generated” and “AI-Assisted.”
An AI-Generated label would apply when generative AI created all or most of a recording’s main creative elements, including a lead vocal or key instrumental performance. AI-Assisted would apply when people remain primarily responsible for the recording but use generative AI for some expressive elements.
The framework is designed for adoption across digital music services and relies on information passed from creators and distributors to streaming platforms. That makes accurate disclosure an important part of the system, particularly when a track passes through several companies before reaching listeners.
Apple Music is taking a related approach. The service has told industry partners that content providers will be required to flag material portions of releases created with AI. Those disclosures, previously handled through Apple’s AI Transparency Tags, are expected to become visible to listeners as “Made With AI” labels later this year.
Spotify’s system addresses a different issue. Its AI Persona badge identifies artist profiles that use an AI-generated, photorealistic identity that could be mistaken for a real person. The badge does not indicate that the music itself was made with AI. Spotify has said music from those personas will generally be excluded from editorial and algorithmic recommendations unless a listener has intentionally engaged with the artist.
The chart rules go further than disclosure by determining whether some AI-related recordings can compete in official rankings at all.
IFPI’s principles require recordings developed with generative AI to use properly authorized AI services, remain substantially human-made, and avoid stream or chart manipulation concerns. The standards also call for AI use to be appropriately disclosed to consumers.
ARIA has now put part of that framework into effect. Wholly AI-generated tracks are ineligible for its charts, while recordings that use generative AI in a supporting role can still qualify if they meet the remaining requirements. ARIA can exclude or remove ineligible recordings and adjust chart positions, and artists may challenge an exclusion through its disputes process.
Those requirements are industry rules rather than new copyright laws. Music companies and chart operators can decide which recordings qualify for their rankings even while courts continue deciding separate questions about whether AI companies may lawfully use copyrighted music to train their models.
That distinction gives the requirement for an “authorized” AI service added significance. A song created with one generative AI platform could face different chart consequences from a similar recording made with a service that has licensing agreements covering the music used to develop its models.
Universal Music Group, Sony Music Entertainment and Warner Music Group were among the record companies that sued AI music companies Suno and Udio in 2024, accusing them of using copyrighted recordings without permission to train generative AI systems. The companies disputed the claims, and Suno and Udio have argued that their use of copyrighted recordings in AI training is legally protected.
The legal and commercial landscape has since shifted. Warner settled its litigation with Suno and Udio and entered licensing partnerships with both companies. Universal also settled with Udio and reached an agreement to develop a licensed AI music platform. Sony has continued litigating against Udio and filed another lawsuit in July accusing the company of infringing more than 30,000 recordings.
Record companies are increasingly distinguishing between AI systems trained through licensing arrangements and services they accuse of building models with copyrighted music without permission.
Australia’s restrictions followed scrutiny of an AI-assisted version of Madonna’s “Like a Prayer” released by DJ Josh Fawaz. The song’s success on Australian radio and the ARIA Charts intensified questions over how AI-created and AI-assisted recordings should be disclosed and counted.
Enforcing the emerging standards remains a challenge because of the volume of music reaching streaming platforms.
Deezer reported in July that fully AI-generated tracks exceeded half of its daily new music uploads at peak levels in June, reaching about 90,000 tracks per day. The platform detects and tags AI-generated music, removes it from algorithmic recommendations, and has said fully synthetic tracks account for only a small percentage of overall listening.
Apple Music has reported a similar divide between supply and consumption. More than a third of music submitted to the service each month is fully AI-generated, according to Apple Music executive Oliver Schusser, while those recordings account for less than 0.5% of listening.
The result is an increasingly fragmented system in which an AI-generated song may be labeled, excluded from recommendations, or barred from a chart depending on the platform, market, and technology used to create it. More chart operators are expected to consider IFPI’s standards as AI-generated music continues to account for a growing share of new streaming uploads.
Universal Music Group and Sony Music Entertainment are pointing to Suno’s licensing deals with Warner Music Group and BMG as they pursue claims that the AI music company used copyrighted recordings without permission to train its models.Suno maintains that its use of copyrighted music for artificial intelligence training is protected... Read More »
AI-generated nude and sexual deepfakes can now lead to criminal charges under federal law and in a growing number of states, as lawmakers move to punish realistic sexual images created or shared without a person's consent.The rules remain a patchwork. Federal law largely targets publication, while states including Florida, Texas... Read More »
A Twitch streamer has sued Twitch and Amazon in a proposed class action, claiming the companies used creators’ videos, chats and other content to train generative artificial intelligence models without their consent or compensation.Warren Pandiscia, a Connecticut creator who has streamed on Twitch for nearly a decade, filed the lawsuit... Read More »
Hollywood and TikTok parent company ByteDance have reached an agreement aimed at strengthening copyright protections across the company’s generative AI products, following months of tension over the use of film and television intellectual property. The Motion Picture Association and ByteDance announced Monday that they have signed a memorandum of understanding... Read More »
Two of the world's biggest record labels are moving to shut down an unauthorized streaming app that's continued operating in Canada despite being pulled from Apple's App Store. Sony Music Group and Universal Music Group are initiating legal proceedings against the developers of Musi, an app that pulls music directly... Read More »
