AI Music Credits and Disclosure

A "credit" is a metadata claim: a statement, attached to a specific recording or composition, that a named person or entity performed a specific role. Credits are not decoration. They feed royalty splits, publishing registrations, search and discovery, and — as of 2025-26 — a growing set of AI-disclosure fields that streaming services show directly to listeners.

For a working producer, getting into AI music production without understanding credits is a common gap: the track sounds finished, but the metadata submitted with it is thin, wrong, or silent on questions a distributor or store will eventually ask.

The roles credits describe

Standard release metadata separates several roles that often overlap in a single bedroom-producer workflow but are tracked independently downstream:

• Performer / artist — the name the release is credited to publicly.
• Producer — the person who shaped the recording: arrangement, sound choices, mix direction.
• Songwriter — wrote lyrics and/or top-line melody; feeds publishing, not just the recording.
• Composer — wrote the underlying musical composition (melody, harmony, arrangement structure), distinct from the sound recording itself.
• Featured / additional performers — anyone who performed audibly on the track but isn't the primary artist, e.g. a guest vocalist or session player.

These roles exist independently of any AI involvement. A track made largely with a text-to-song platform still needs a performer credit and, per most distributor terms, a human rights-holder behind the release. Where AI changes the picture is in a newer, separate field: disclosure.

Where credits live: metadata, not paperwork

Credits are entered once, at upload, as structured metadata — not a separate document. They travel with the release through the distributor to each streaming service via standard music-industry data messages, then surface on the store's own credits panel (where a service chooses to show one). This matters because it means metadata differs between services: a field DistroKid captures at upload does not automatically appear the same way, in the same place, or at all, on every platform the release reaches. Two listeners on two different services can see two different credit panels for the identical file.

AI disclosure is a narrower, newer idea than crediting: a flag that says whether AI generated part of the work, independent of who is credited as artist or producer. Crediting answers "who made this and in what role"; disclosure answers "was any of this generated, and how much". A track can have complete, accurate performer and producer credits and still need a separate answer to the disclosure question.

DistroKid's help documentation defines its own AI Credits field this way: it discloses when AI generated part of a track — lyrics, vocals, or instrumental performance — and that disclosure is then shown to listeners on the streaming services that currently display it, which DistroKid names as Spotify and Apple Music (DistroKid, AI Credits). Crucially, DistroKid's own guidance draws a line at what does not require this disclosure: AI used as a production tool — pitch correction or auto-tune, AI-assisted mixing or mastering, AI-assisted workflows generally — is treated differently from AI that generated the audio, lyrics or composition itself.

That distinction is why "AI-generated" and "AI-assisted" are not interchangeable labels, and why an industry labelling proposal has formed around exactly that split (covered below). If your workflow is closer to the assisted end — using a generation engine for a sketch, then rebuilding drums, bass and arrangement by hand — read AI-generated vs AI-assisted music for how that distinction is usually drawn before you fill in any disclosure field.

Two practical reasons this matters for producers, beyond ticking a box correctly:

1. Accuracy protects the release. Distributor terms generally require truthful disclosure; getting it wrong is a metadata-accuracy problem, separate from — but related to — content policy at the distributor or store.
2. Disclosure fields feed eligibility questions elsewhere. Chart bodies and some platform policies are starting to ask whether a recording is "substantially human made" (see IFPI's chart principles below); accurate disclosure metadata is the raw material those checks will eventually run on.

Because disclosure and eligibility questions are being built on top of metadata you supply once, at release time, it is worth keeping your own production notes rather than reconstructing the answer months later. A simple session log — which stems came from a generation engine, which elements were replaced or re-recorded, which vocal takes are human performances, which mix and master passes used AI-assisted tools — does three things: it makes filling in an AI Credits field at upload fast and accurate, it gives you a defensible answer if a distributor or platform asks a follow-up question, and it protects you if collaborators or a label later need to know exactly what's in the file.

This is a documentation habit, not a legal opinion. Copyright outcomes for AI-involved work depend on the law of the country you're releasing under and on the extent of human authorship; the position is not settled and varies by jurisdiction, so treat any ownership question as one for a copyright specialist rather than a streaming service's FAQ page — see MuzeMe's own copyright information for how that applies to tracks built with an AI production platform.

This section covers policies with a published, dated source you can check yourself. Each is a distributor or platform's own documented feature, not an industry-wide rule.

DistroKid: AI Credits field

DistroKid's upload flow asks directly whether any part of a release was generated by AI, with granular options: the lyrics, the music (melody), all of the audio, or part of the audio. Choosing "all of the audio" triggers a follow-up asking whether the artist name represents a human or an AI persona. Credits can be added or updated after the initial upload (DistroKid, How to Fill Out AI Credits). DistroKid's companion article confirms this disclosure is shown to listeners on Spotify and Apple Music currently, and explicitly excludes AI-assisted mixing, mastering and pitch correction from the requirement (DistroKid, AI Credits).

Spotify: displaying disclosure

Spotify's own announcement describes AI disclosure in credits as one of three protection measures it is rolling out, alongside impersonation enforcement and a music spam filter. Spotify states the disclosure mechanism is an industry standard being developed through DDEX (the metadata standards body for music messaging) rather than a Spotify-only field, and frames AI use as a spectrum rather than a single yes/no flag (Spotify Newsroom, 25 Sep 2025). In the twelve months before that announcement, Spotify says it removed over 75 million spammy tracks — separate from disclosure, but part of the same trust push.

Deezer: AI tagging on-platform

Deezer runs its own detection system rather than relying solely on uploader disclosure. Albums containing fully AI-generated tracks are tagged; tagged tracks are removed from algorithmic recommendations and editorial playlists, though they remain available to stream. Deezer reports its detection tool identifies fully AI-generated songs and separately says it has found a high proportion of streams on those tracks to be fraudulent (Deezer says up to 85% of streams on fully AI-generated songs were identified as fraudulent as of January 2026), which it demonetises (Deezer Newsroom, 20 Jun 2025). This is a platform-side detection and tagging system, distinct from an uploader-filled credit field.

Beatport: ingestion-time tagging

Beatport's updated content guidelines, reported on its own editorial site, tag tracks made with AI assistance during ingestion for curation transparency, provided the finished track remains majority human-made; fully or majority AI-generated tracks are not permitted and are withheld, with rightsholders notified. Beatport pairs this with a detection partnership (Beatdapp) aimed at flagging AI-generated material before it reaches the store (Beatportal, 12 Aug 2026). This is a tagging-at-ingestion mechanism tied to content policy, not a disclosure field a listener sees on every track.

The items below are proposals, metadata standards or delivery specifications. None is a law, and none is a uniform rule every distributor or store has adopted. Keep three things apart: credits (who contributed and in what role: artist, performer, composer, songwriter, producer, recording, mixing and mastering engineer), AI disclosure (information about AI use in making the recording), and metadata (the data carried through the supply chain, which can hold either).

IFPI: voluntary AI-Generated / AI-Assisted labels

In July 2026, IFPI announced a labelling programme developed jointly with RIAA, A2IM, WIN, IMPALA, The Grammys, SAG-AFTRA and the Human Artistry Campaign, proposing a unified, voluntary approach to track labelling that distinguishes "AI-Generated" from "AI-Assisted" recordings. The announcement frames this as intended for broad adoption across digital music services and explicitly designed to evolve over time (IFPI, 10 Jul 2026). No source confirms every major store has implemented this two-tier label; it is a proposed industry standard.

DDEX: AI signalling in the supply chain

DDEX is the standards body behind the metadata messages labels and distributors use to deliver releases to services. DDEX's own AI initiative page says its release standard (ERN) has been updated so a record company or distributor can signal to a DSP, at a very simple level for now, the amount of AI involved in creating a sound recording or music video, and that work continues on how AI involvement is communicated through the supply chain and ultimately to consumers (DDEX, Other Initiatives). Spotify says its AI credits build on a DDEX standard (Spotify Newsroom, 25 Sep 2025). A DDEX field is a capability of the metadata standard. It is not a platform requirement, a distributor requirement or a legal obligation, and it does not mean every service shows it to listeners.

Apple: AI transparency tag for music video singles

Apple's own delivery documentation, Apple Music Specification 5.3.26 (April 2026), added an <ai_transparencies> tag for disclosing when AI was used to generate a material portion of a music video single (Apple Music Specification). It is delivery metadata supplied by labels and distributors. It does not by itself establish an Apple requirement to disclose AI for every song, and it is not a credit category. Separately, DistroKid says Apple Music is one of the services that displays its AI Credits disclosure (DistroKid, AI Credits).

IFPI: chart eligibility principles

Separately from labelling, IFPI announced principles in July 2026 for when a recording developed using generative AI can be included in official charts: the AI service must be authorised and lawful, the recording substantially human made, and there must be no manipulation concerns, alongside compliance with applicable law, the AI service's terms and appropriate downstream signalling. IFPI is rolling these out across the official charts in its own network (IFPI, 30 Jul 2026). They are IFPI's chart principles, not a legal standard and not a rule for every chart worldwide.

Because these two categories get blurred in casual coverage, here's the same information side by side. "Documented" means there is a published, dated policy or feature you can point to; "developing" means an announcement of direction or a proposal without a confirmed, uniform rollout.

The practical read: fill in whatever disclosure field a distributor gives you accurately, expect differences between services on how — or whether — that shows up to listeners, and don't assume a voluntary proposal is already a rule your release must satisfy. For what's actually enforced at upload and ingestion rather than at the credits/labelling stage, see AI music distribution.