AI Music Disclosure: what to declare on your release - and what it changes

Generative AI is capable of creating music that’s almost indiscernible from human-generated music. The more it advances, the more some listeners are focusing on consciously avoiding listening to AI-generated music. And streaming platforms have caught on.

Starting with Deezer, many DSPs have followed suit. Now, as of October 2026, Spotify, Apple Music, YouTube Music, Qobuz, Tidal and Bandcamp have all followed in Deezer’s footsteps by displaying badges or other visual indicators of either AI-generated artist personas or tracks that have been detected as or disclosed as AI-generated.

Distributors, like Identity Music, are rolling out AI scanning and self-disclosure labels in release metadata - but what counts as AI-generated, and will disclosing use of AI harm your streams?

‍

What is AI disclosure in music metadata?

When your music is distributed to streaming platforms, download stores, or social media platforms, it’s sent with a variety of metadata via DDEX delivery feeds.

DDEX bundles all relevant data about your release into a format that can be automatically ingested by DSPs. Part of that data is your metadata - including release title, track title(s), artist(s), genre, publishing credits, instrumentation, samples disclosure and, most recently, AI disclosure.

AI disclosure in music metadata is an entirely optional field. It’s not (yet) something that any DSPs require to be completed or proven in any way, but it is becoming ‘best practice’ to declare use of AI if you have done so. In most distribution platforms, it’s a field very similar to the Beats/Samples Usage disclosure field: a tick-box or yes/no question that can be answered positively or negatively by the uploader.

‍

Does declaring AI usage affect royalties or reach?

AI disclosure is a very new thing for most DSPs, so being able to say with 100% confidence whether or not it affects your music career trajectory over the long term isn’t yet possible. What we can say, however, is that platforms are beginning to treat self-disclosed or DSP-detected fully AI-generated tracks differently from tracks that are human-made.

‍

How Spotify treats AI-generated music

Spotify has taken a broader stance than some other DSPs and instead of banning AI-generated music, Spotify places AI-generated music alongside ‘spam’ content in a category of music that has been excluded from recommendation algorithms. The music remains live on the platform and is still eligible for royalties like any human-made track, but isn’t included in algorithmic contexts like Radio, DJ, or other recommendation-driven placements.

Spotify announced this measure in September 2025 and has reportedly removed 75 million ‘spammy’ tracks from the platform altogether and has taken 50 million AI-generated ‘slop and spam’ tracks out of recommendation algorithms in the first year.

Spotify uses a combination of self-disclosure in metadata as well as listener engagement, upload patterns (it’s unrealistic for a human to be creating hundreds of tracks a month, for example), and data from third-party AI detection tools. Tracks that are identified as being AI-generated are removed from recommendations, thus limiting reach and likely impacting earned royalties too.

You can find out more about Spotify’s approach to ‘putting human artistry first’ here.

‍

How Apple Music treats AI-generated music

While not as far progressed as Spotify, Apple Music recognises the threat posed by AI-generated music and is beginning to implement similar measures.

All we know so far is that Apple Music is rolling out ‘Made With AI’ labels applied to any content that has been disclosed as being AI-generated. As of writing, there is no concrete plan to introduce AI detection, unlike Spotify, and instead Apple Music’s ‘Made With AI’ label will rely on self-disclosure in metadata when an artist or label uploads music to their distributor.

Currently, there are also no plans to demonetise or restrict the reach of AI-generated music. Instead, Apple Music’s goal is to provide listeners with transparency and choice. It is yet to be seen whether algorithmic restrictions may come into place in future updates.

Find out more about Apple Music’s intentions for the ‘Made With AI’ label here.

‍

How Amazon Music treats AI-generated music

As of October 2026, Amazon Music has not introduced a policy on how AI-generated music will be labelled or treated by the platform. Amazon claims to be ‘part of those conversations’ when it comes to handling AI-generated music, but there has been no official update as to whether or how AI-generated music will be labelled or restricted by Amazon Music.

‍

How Deezer treats AI-generated music

Deezer has been building and refining its own AI-music detector for a few years now and has been reporting its AI stats since January 2025. AI-generated tracks accounted for 10% of daily uploads then, rising to more than 50% in June 2026.

Deezer is actively detecting and labelling fully AI-generated tracks and, similarly to Spotify, excluding those tracks from algorithmic or editorial placements. More recently, Deezer has started to proactively remove tracks that are fully AI-generated if they’ve been involved in suspected streaming fraud or haven’t been streamed for six months or more.

While Deezer isn’t removing every AI-generated track it detects, it is restricting their reach and paying close attention to their engagement levels to spot early signs of streaming fraud. 

You can find out more about Deezer’s stance on AI here.

‍

How YouTube treats AI-generated music

YouTube hasn’t made any public statements specifically about AI music, but it does already require content creators to disclose when AI has been used in any part of their videos, including music. At the time of writing, there are no algorithmic restrictions on AI-generated music or content, merely a label added to AI content for viewer transparency.

‍

How Tencent treats AI-generated music

Tencent’s public stance on AI-generated music concerns its copyright compliance. Tencent processed takedowns of over a quarter of a million ‘policy-violating’ tracks in 2025, many of which were likely to be AI-generated soundalikes or covers of copyrighted material.

Using its own AI detection models, Tencent reports to identify and tag content containing AI-generated material. It does not currently confirm any restrictions on royalties earned by that content or any algorithmic or editorial reach implications, but Tencent does label AI-generated music for listener visibility.

‍

How SoundCloud treats AI-generated music

SoundCloud does not allow AI-generated content that impersonates other artists or their voices. Outside of that restriction, SoundCloud has publicly said little about any restrictions to self-declared or detected AI-generated music.

‍

How Tidal treats AI-generated music

While later to follow suit with the likes of Spotify when it comes to handling AI-generated music, Tidal has issued rules about AI-generated content as of June 2026. Music identified (either by self-disclosure or detection) as fully AI-generated will not be monetised on Tidal at all. Tidal has stated that distributors are responsible for identifying and labelling AI-generated music before distributing to Tidal, making use of the self-disclosed AI usage label in track metadata.

Tidal has stated that it will look to roll out distinct labelling for fully AI-generated music and ‘substantially AI-generated’ music, with the responsibility for labelling content lying with artists, labels and distributors, not just Tidal.

You can read Tidal’s AI Policy in full here.

‍

Distributing AI-generated music

The music industry as a whole is somewhat divided when it comes to the risk vs benefit of AI in music. While some distributors remain open to all, regardless of whether your music is human-created or entirely AI-generated, others, like us at Identity Music, have drawn a line in the sand when it comes to distributing fully AI-generated music.

Before we look at distribution restrictions placed on AI music, we must first note what we actually mean by AI-generated music.

Like most things, AI’s influence on music comes on a spectrum. 

At one end, there’s fully AI-generated music: where a human gives an AI model a prompt and a track is generated based on that prompt. There’s no human influence besides the words and directions given in the prompt. 

Then, at the other end, there’s human-created music that uses AI-powered assistive tools in part of the process: humans have written and performed the track, then used AI-powered mixing or mastering tools before releasing. Without the human creating the concept and performing the track, the track would not exist.

And then there’s the middle of the spectrum, where tracks might be partly human-written and human-performed, partly AI-written and AI-performed. It’s this middle part of the spectrum that’s the most difficult to reliably identify, because there are many different elements that could be generated by AI, assisted by AI, or completely human-created.

At Identity Music, we believe first-and-foremost in the power of human creation. Whilst we acknowledge that AI has its place, we don’t believe that generative AI in the music industry is a net positive. For that reason, we draw a distinction between fully AI-generated music, which we don’t distribute, partially AI-generated music, which we may distribute, and AI-assisted music, which we do distribute.

You can learn more about our stance on distributing AI music here.

‍