Registering AI-Assisted Music with the U.S. Copyright Office: What Creators Need to Know
Failing to disclose AI-generated elements in a copyright application can put the entire registration at risk.
Disclaimer: This article is for general informational purposes only and does not replace legal advice for specific situations.
Example: A producer uses Suno to generate an instrumental track, writes original lyrics over it, and files a copyright registration application without disclosing the AI-generated instrumental. The registration is granted. Months later, during a licensing dispute, the omission is discovered and the registration's validity is challenged in court [1].
Disclosure is not a technicality. It is a condition that determines whether a registration holds up when it matters most.
TL;DR
Music creators using AI tools can register copyright in their AI-assisted works — but only for the portions they actually authored. The U.S. Copyright Office requires disclosure of AI-generated content and limits protection to human-authored expression. Under the human authorship standard confirmed by the USCO's Part 2 Report (January 2025) and the D.C. Circuit's ruling in Thaler v. Perlmutter (March 2025), prompts alone are insufficient to establish authorship. What matters is whether a human exercised sufficient creative control over the expressive elements of the final work.
The human authorship threshold: What "sufficient control" means
The Copyright Act of 1976 has always required human authorship as a precondition for copyright protection [2]. For AI-assisted music, the operative question is not whether AI was used — it is whether the human creator exercised sufficient creative control over what the audience ultimately hears.
The USCO's January 2025 Part 2 Report on Copyright and Artificial Intelligence synthesizes the current framework into four categories of human contribution, each carrying different implications for registrability [2]:
| Category | What it covers | Registrable? |
|---|---|---|
| AI as a creative tool | Using AI for ideation, refinement, or enhancement while the human determines expressive output | Yes — human authorship is unaffected |
| Prompts only | Entering text or audio prompts and accepting the AI's output as-is | No — prompts do not provide sufficient human control over expressive elements |
| Expressive inputs | Supplying original human-authored material (e.g., a melody, lyrics, vocal performance) that remains perceptible in the AI output | Yes — the human-authored portion is protectable |
| Modification or arrangement | Selecting, arranging, substantially editing, or re-recording AI-generated material in a sufficiently creative way | Yes — the human-authored modifications or arrangement may be protectable, but the underlying AI-generated material itself remains unprotected |
The dividing line in each case is whether the human — not the AI system — determined the expressive elements of the final work [2]. Prompt engineering, even with highly detailed or iteratively refined prompts, falls below this threshold under current technology: the AI system remains responsible for translating the user's instructions into fixed expression [2].
The D.C. Circuit's March 18, 2025 ruling in Thaler v. Perlmutter reinforced this at the judicial level, holding that human authorship is required as "a matter of statutory law" under the Copyright Act [4]. The Supreme Court declined to review the case in March 2026, leaving that ruling as controlling precedent [9].
See more: "What Is Music Copyright?" | "What is Not Protected in Music Copyright?"
The Disclosure Obligation: What the Application Requires
When an AI-assisted music work contains more than a de minimis amount of AI-generated material, the Standard Application on copyright.gov must be completed with the following disclosures [1]:
Author Created field — Describe only the material the human author contributed. For AI-assisted music, this means describing the human-authored elements specifically: original lyrics, melodic arrangement, vocal performance, selection and sequencing of AI-generated elements, or substantive edits to AI-generated instrumentation. Do not describe AI-generated content here.
Material Excluded / Limitation of Claim field — Explicitly disclaim the AI-generated portions. Examples of acceptable language include: "Musical composition generated by artificial intelligence," "Instrumental tracks generated by AI (Suno)," or "Sound recording elements generated by AI production tools" [1][11]. Describing the disclaimed material with sufficient specificity is important — disclaiming too broadly may forfeit rights the creator could have claimed; disclaiming too narrowly may result in examiner correspondence or leave the registration vulnerable to challenge.
Note to Copyright Office (optional but recommended) — Use this field to explain how AI was used and what the human author contributed. A well-constructed note preempts common examiner questions by clarifying the relationship between the AI's contribution and the human's creative decisions [11].
One filing requirement worth noting: AI-assisted works must use the Standard Application, not the Single Application. The Standard Application provides access to the Limitation of Claim fields necessary to disclaim AI-generated material [11].
See more: "How to Protect Your Music Copyright?"
Consequences of Non-Disclosure
Failing to disclose AI-generated material in a copyright application carries two distinct categories of risk [1][6]:
First, the USCO may cancel the registration on its own initiative if it becomes aware that information essential to its evaluation of registrability was omitted. This can occur through examiner review, third-party notification, or — as in the Zarya of the Dawn incident — public disclosure through social media [6].
Second, in litigation, an adverse party may invoke Section 411(b) of U.S. Copyright Law to argue that the registration should be disregarded because the applicant knowingly provided inaccurate information and the accurate information would have caused the Office to refuse registration [1][6].
The disclosure obligation is also retroactive for existing registrations. Creators who registered AI-assisted works before the 2023 guidance — or failed to disclose at the time — may correct the record by filing a supplementary registration, amending the "Author Created" and "Material Excluded" fields accordingly [1].
See more: "What is 'Poor Man's Copyright'?"
Practical Scenarios for Music Creators
The following scenarios illustrate how the USCO framework applies in common AI-assisted music workflows:
| Scenario | Human contribution | USCO outcome (likely) |
|---|---|---|
| Prompt only → accept AI output as-is | None beyond prompting | Not registrable — no human authorship over expressive elements [2] |
| Prompt → substantial lyric rewrite + vocal recording | Original lyrics + performance | Registrable for lyrics and sound recording; disclaim AI-generated musical elements [1][2] |
| Prompt → generate multiple AI-generated elements → creatively select, coordinate, and arrange into a cohesive final track | Creative selection and arrangement | May be registrable for selection/arrangement if sufficiently creative; disclaim underlying AI-generated musical material [2][5] |
| Human-composed melody → fed into AI tool as expressive input → output retains that melody perceptibly | Human-authored melody perceptible in output | Registrable for the portion reflecting the human-authored melody [2][8] |
| Human melody fed into AI → AI transforms it into unrecognizable output | Human authorship not perceptible in final output | Not registrable — cf. SURYAST, where the AI tool's processing severed the link between human input and output [7][8] |
The SURYAST decision (December 2023) is instructive for musicians who use their own compositions as AI inputs: if the AI's processing transforms the input to the point where the human-authored material is no longer distinguishable in the output, the registration claim fails — even if the human contributed original material at the start [7].
Conclusion
For AI-assisted music, registration protection is available — but its scope is defined by what a human actually authored, not by the effort spent prompting. The practical implication is straightforward: frame the registration claim around the lyrics written, the arrangement crafted, the performance recorded, and the substantive creative decisions that shaped what the audience hears.
Document those contributions during production, not after. A registration that accurately reflects the human-AI boundary is far more defensible than one that overstates human authorship and collapses under scrutiny.
REFERENCES
[1] U.S. Copyright Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 88 Fed. Reg. 16,190 (Mar. 16, 2023);
[2] U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (Jan. 29, 2025);
[3] U.S. Copyright Office, AI Initiative — copyright.gov/ai (last visited May 2026);
[4] Thaler v. Perlmutter, No. 23-5233 (D.C. Cir. Mar. 18, 2025);
[5] Congressional Research Service, Generative Artificial Intelligence and Copyright Law (LSB10922, updated 2025);
[6] Finnegan, Henderson, Farabow, Garrett & Dunner, LLP, "When Registering Works Incorporating AI-Generated Material, It Pays to Be Truthful" (2025);
[7] Sterne Kessler, "The U.S. Copyright Office's Position on the Copyrightability of Works Made with the Assistance of Generative AI (Part Two)" (2024);
[8] Skadden, Arps, Slate, Meagher & Flom LLP, "Copyright Office Publishes Report on Copyrightability of AI-Generated Materials" (Feb. 4, 2025);
[9] Mayer Brown, "Supreme Court Denies Review in AI Authorship Case" (Mar. 2026);
[10] Copyright Alliance, "How to Register AI-Generated Works with the Copyright Office" (2023);
[11] RightsDocket, "Limitation of Claim for AI Music: The Complete Guide to USCO Copyright Registration" (reviewed Apr. 3, 2026).