🏛️ Issues and Advocacy Training
Courses › AI for Voice and Music

Lesson 8 of 9 · about 18 minutes

Copyright, Platform Rules and Disclosure for AI Audio

You'll be able to

  • Sort the parts of an AI-assisted audio project into human-authored and AI-generated
  • Summarize the status of each label lawsuit against Suno and Udio as of October 2026
  • Apply Spotify's and YouTube's AI rules to a release or upload
  • Recognize clear disclosure in show notes, descriptions, metadata and spoken credits

Two different questions hide inside "can I use this AI song?" One is copyright: does anyone own it, and could you? The other is contract: what does the service's terms of use let you do with what it generated? A track can be yours to sell under the contract and still be unprotected by copyright, which means you can use it and so can anyone who copies it.

What you can own

The U.S. Copyright Office's report on copyrightability, released January 29, 2025, concluded that generative AI output is protected only where a human author determined sufficient expressive elements. That can include a human-authored work that is perceptible in the output, or a human's creative arrangement or modification of the output. It does not include the mere provision of prompts. The module on AI images and video walks through that report and the Supreme Court's March 2026 refusal to hear Thaler v. Perlmutter, which left the human-authorship rule standing.

Music splits more cleanly than images, because a song has separable layers.

Layer of a MusicStudio songWho made itProtectable by you?
Lyrics you wroteYouYes
Lyrics from "Write with AI"The modelNo, unless you rewrote them substantially
Melody, chords, arrangement, vocal performanceThe modelNo
Your edits: a restructure, a new bridge you played, your own guitar track added over itYouYour contributions, yes
Style promptYouNot as part of the song
Narration in your own recorded voiceYouYes
Narration in a synthetic voice reading your scriptYou wrote it; the model performed itThe script, yes; the synthetic performance, no

Suno's own help center says much the same in plain terms: music made 100 percent with AI would not qualify for U.S. copyright, writing the prompt does not make you the author, and if you wrote the lyrics, you own the lyrics.

If you register, the Copyright Office's 2023 registration guidance requires you to identify AI-generated material and exclude it from your claim. Claim the lyrics and your own additions. Disclaim the AI music. Your project log, the kind lesson 3 showed Dialogue Editor keeping automatically, is how you prove which is which.

What the contract lets you do

ServiceFree tier, as of October 2026Paid tier
SunoSuno owns the songs; personal, non-commercial use; up to 7 lifetime trial downloadsYou own songs made while subscribed and keep commercial rights after you cancel; monthly download limits
Mureka (behind MusicStudio)Daily free credits for a few songs on an older modelFAQ says subscriptions include full commercial usage rights; downloads stop if the subscription lapses
ElevenLabs (Voice Studio, Dialogue Editor)No commercial license; published content must credit ElevenLabsCommercial license, except for beta services
UdioDownloads off on every plan since late 2025, pending a licensed relaunch, per a September 2026 trade summary

One caution for Jeff's tools in particular. MusicStudio and Voice Studio call Mureka's and ElevenLabs' APIs, and API terms can differ from the consumer app terms quoted in help centers. The tool pages do not state licensing terms. For anything sponsored or sold, read the provider's current API or platform terms for the account that pays.

The lawsuits, case by case

In June 2024 the three major label groups sued both Suno and Udio. Here is where each claim stood in the latest reports found for this lesson.

PlaintiffAgainst Suno (D. Mass.)Against Udio (S.D.N.Y.)
Universal Music GroupPending. A stream-ripping claim under the DMCA's anti-circumvention section was allowed in August 2026; fact discovery was set to close September 30, 2026, with summary judgment motions on fair use expectedSettled October 2025, with licensing and a planned joint platform
Sony MusicPending, alongside UniversalPending. Sony filed a second suit in the same court on July 20, 2026, adding a YouTube stream-ripping claim, after the judge would not let it amend the first
Warner Music GroupSettled November 25, 2025; Suno to build licensed models and limit downloads; Warner artists can opt in to likeness useSettled November 19, 2025

Each side's strongest point deserves a fair hearing. The labels point to Suno's admission that it trained on tens of millions of recordings, including audio taken from YouTube, to build a product that competes with recorded music. Suno argues the copying is an internal step toward non-infringing new music, notes the labels do not appear to claim its outputs copy their recordings, and accuses the labels of anticompetitive conduct. The Copyright Office's pre-publication training report, as lesson 5 described, leans toward the labels on market-competing commercial uses but stresses that each case turns on its facts. No court had ruled on fair use in these cases as of October 2026. Check PACER or CourtListener before you repeat any status here.

Platform rules

Spotify

In September 2025 Spotify announced three measures, without banning AI music. It restricts vocal impersonation, allowing a cloned or imitated artist voice only with that artist's authorization. It announced a music spam filter, to roll out that fall, aimed at mass uploads, duplicates, SEO tricks and artificially short tracks, after saying it had removed over 75 million spam tracks in 12 months. And it backs a DDEX industry standard for AI credits covering vocals, instrumentation and post-production, submitted by labels and distributors. Spotify says disclosure will not cause penalties or down-ranking; a later update says tens of thousands of AI credits arrive daily, and that a missing credit does not mean AI was not used.

YouTube

YouTube requires creators to disclose realistic content that is generated or meaningfully altered, including music that is the main focus of a video, and its list of examples includes "AI generated music." It says cloning your own voice for a voiceover, using AI to help write a script, and audio repair do not need disclosure. The label may appear in the player for photorealistic content and in the expanded description otherwise, and YouTube says disclosure does not limit audience or monetization eligibility.

Separately, YouTube renamed its repetitious content monetization policy "inauthentic content" on July 15, 2025. AI content made with generic templates that gives the impression of mass production, without the creator's own perspective, is not eligible for monetization. A channel of 300 AI songs over stock loops is exactly what that rule describes.

Disclosure that travels

Audio gets separated from its context constantly: a clip ripped from a video, an episode played in a car with no show notes on screen. Disclosure that survives sits in more than one place, and at least once in the audio itself when the synthetic element carries meaning.

WhereExample wording
Spoken, in the piece"Our narrator's voice is AI-generated."
Podcast show notes"Narration: synthetic voice (ElevenLabs), script by the host. Theme music: AI-generated (Mureka), edited by the host."
Video description and YouTube's disclosure toggleSame text, plus "Yes" on the altered or synthetic content setting where it applies
File metadataID3 comment or description field: "AI voice and AI music, see show notes"
Streaming releaseAI credits through your distributor, where offered
Dramatized scenes"The voices in this scene are synthetic. The people are invented."

Worked example: a release checklist with numbers

A local group releases a 4-song EP: each song has the member's own lyrics and AI music from MusicStudio, with one member's real guitar added on two songs.

  • Copyright claim: lyrics on 4 songs, plus guitar parts on 2. AI music disclaimed on all 4.
  • Contract: generated on a paid Mureka account, so commercial use is allowed under the FAQ; confirm the API terms.
  • Spotify: AI credits for instrumentation and vocals on all 4 tracks through the distributor.
  • YouTube lyric videos: disclosure toggle set to Yes on all 4 (AI-generated music is the main focus).
  • Records: generation IDs, prompts, dates and the 30-day download deadline noted for each song.

Reporting angle

  • When a city, school district or campaign uses AI music or a synthetic narrator in a paid ad or video, ask which service made it, on what plan, and whether anyone checked the commercial terms. Public bodies can be asked for the invoice under public records law.
  • For a story on a local musician releasing AI-assisted work, ask what they wrote and what the model made, and whether they registered anything with the Copyright Office. Registrations are searchable in the Office's public records system.
  • When the Suno or Udio cases produce a fair use ruling, read the opinion itself before the press releases. Both sides will claim victory on partial rulings.

Research angle

Common mix-ups

  • Claim: If I paid for the plan, I hold the copyright. What the evidence shows: A plan grants contractual rights. Copyright needs human authorship, and the AI parts have none.
  • Claim: Spotify bans AI music. What the evidence shows: Spotify targets impersonation and spam and supports voluntary AI credits. AI music itself is allowed.
  • Claim: Disclosure hurts reach on YouTube. What the evidence shows: YouTube says disclosure does not limit audience or monetization eligibility. Mass-produced template content is what loses monetization.

Check yourself

1. You wrote the lyrics; MusicStudio generated the melody, arrangement and vocal. Under the Copyright Office's January 2025 report, what can you claim?

2. As of September 2026, which plaintiffs were still suing Suno in federal court in Massachusetts?

3. Which of these does YouTube's disclosure policy list as needing a disclosure?

4. What does Spotify say about tracks that disclose AI use through the new AI credits standard?

5. Since July 15, 2025, YouTube's monetization policy calls mass-produced, repetitive channels what?

6. Suno's help center says songs made on the free plan are owned by whom?

Terms in this lesson

Human authorship
The U.S. requirement that a work have a human author to be protected by copyright.
Disclaim (in registration)
To identify and exclude AI-generated material from what you claim when registering a copyright.
DDEX
An industry standards group for music metadata. Spotify supports a DDEX standard for AI credits.
Impersonation
On streaming platforms, releasing music with a cloned or imitated voice of a real artist without permission.
Provenance
The documented history of how a piece of media was made and changed.

Full glossary

Sources