AI, Copyright, and Rulemaking

How the U.S. Copyright Office, public comment, and emerging EU analysis are shaping the rules for AI‑generated works.

1. The Emerging Policy Landscape

The rapid diffusion of generative‑AI systems has forced copyright policymakers to revisit long‑standing doctrines of authorship, originality, and registration. In 2023 the United States Copyright Office announced a formal study of “the copyright law and policy issues raised by artificial intelligence (“AI”) systems” ([Artificial Intelligence and Copyright], 2024). The study is not a final rule but a fact‑finding exercise that will inform possible legislative amendments or regulatory guidance.

At the same time, the Office issued a separate “Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence” ([Copyright Registration Guidance], 2024) that clarifies how examiners will treat AI‑assisted submissions today. Together, the inquiry and the guidance create a two‑track process: (1) a public‑comment‑driven fact‑finding phase, and (2) an interim policy statement that shapes day‑to‑day registration practice.

Internationally, the European Union is conducting its own doctrinal analysis. A 2024 article in GRURRR asks whether “AI‑assisted output” meets the EU’s requirements for a protectable “work” ([Copyright and Artificial Creation], 2024). These parallel efforts illustrate that rulemaking on AI and copyright is happening on multiple fronts, and that stakeholders must monitor both U.S. and EU developments.

2. The Copyright Office’s Notice of Inquiry

The formal Notice of Inquiry (NOI) appeared in the Federal Register as “Artificial Intelligence and Copyright” ([Artificial Intelligence and Copyright], 2024). Its purpose is to solicit information on several core questions:

Whether AI‑generated material can satisfy the statutory “original work of authorship” requirement.
How the concept of “human authorship” should be interpreted when an AI system contributes substantive expressive choices.
What registration procedures, if any, need to be adjusted to capture AI‑assisted works.

The NOI also requests data on the volume of AI‑related registration requests, the types of AI tools being used, and any observed patterns of infringement. The Office explicitly states that the inquiry is “pre‑legislative” and that the findings may lead to rulemaking, legislative proposals, or a combination of both. Because the NOI is a public document, any party can submit written comments that become part of the official record.

3. Stakeholder Input: What the Comments Reveal

3.1 Academic Perspectives

Matthew Sag, Pamela Samuelson, and Christopher Jon Sprigman submitted a joint comment titled “Comments In Response To The Copyright Office's Notice Of Inquiry On Artificial Intelligence And Copyright” ([Comments In Response To The Copyright Office's Notice Of Inquiry On Artificial Intelligence And Copyright], 2024). Their analysis emphasizes three points that have repeatedly surfaced in scholarly debate:

Human Authorship as a Threshold – The authors argue that the statutory language still requires a human author, but they note that “creative direction” provided by a human can satisfy the originality requirement even when the AI produces the final expressive output.
Transparency in Disclosure – They recommend that registrants disclose the extent of AI involvement in a standardized field on the registration form, to give downstream users clear notice.
Policy Flexibility – The comment warns against a “one‑size‑fits‑all” rule, suggesting a tiered approach that distinguishes between AI‑assisted works (human‑driven) and fully autonomous AI‑generated works.

These positions are grounded in the authors’ extensive research on copyright theory and have been cited in subsequent policy discussions.

3.2 Industry and Practitioner Viewpoints

Michael D. Murray offered a separate comment titled “Comment — Re: Notice of Inquiry on Copyright and Artificial Intelligence” ([Comment — Re: Notice of Inquiry on Copyright and Artificial Intelligence], 2024). Murray’s submission reflects concerns common among practitioners and technology companies:

Clarity for Registration – He stresses the need for a clear, binary question on the registration form (“Did a human author create the work?”) to avoid ambiguous determinations.
Risk Management – Murray points out that without definitive guidance, creators risk inadvertent infringement when training AI on copyrighted material.
International Consistency – He urges the Office to consider alignment with emerging EU standards to reduce compliance complexity for multinational firms.

Both comments, while differing in emphasis, converge on the need for transparent disclosure and a nuanced definition of authorship.

3.3 Gaps Identified by Commenters

Across the public record, commenters note several areas where the NOI provides limited direction:

Derivative Works – How to treat AI‑generated outputs that are based on pre‑existing copyrighted works.
Moral Rights – Whether AI‑generated works can trigger moral‑right claims under the Visual Artists Rights Act.
Enforcement – Practical mechanisms for detecting AI‑generated infringement.

These gaps signal where future rulemaking may focus.

4. The Office’s Draft Guidance on AI‑Generated Material

In August 2024 the Copyright Office released a “Statement of Policy” titled “Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence” ([Copyright Registration Guidance], 2024). The guidance does not create new law but clarifies examiner practice for the interim period. Key takeaways include:

Human Authorship Requirement – Registrations will be accepted only when a human author can be identified as the “originating author” of the work. AI‑generated elements may be included, but the work must contain a “substantial amount of human‑created expression.”
Disclosure Requirement – Applicants must complete a new “AI‑Generated Content” checkbox and may attach a brief description of the AI tool used, the prompts given, and the extent of AI contribution.
No Separate Copyright for AI Output – The Office reiterates that “the output of an AI system is not itself eligible for copyright” unless a human author meets the originality threshold.

The guidance also notes that the Office will continue to collect data on AI‑related submissions to inform future rulemaking. While the policy is advisory, it already shapes how registrars evaluate AI‑assisted works.

5. Comparative Lens: EU Approach to AI‑Assisted Works

The EU’s legal framework has long required a “personal intellectual contribution” for a work to be protected. The 2024 GRURRR article “Copyright and Artificial Creation: Does EU Copyright Law Protect AI‑Assisted Output?” ([Copyright and Artificial Creation], 2024) conducts a doctrinal analysis of this requirement.

The authors conclude that EU law can accommodate AI‑assisted works if the human contributor exercises “creative freedom” in selecting prompts, configuring parameters, or curating the output. They argue that the “originality” test is satisfied when the human’s contribution is more than a trivial selection. However, the article also warns that fully autonomous AI output—where no human makes a creative choice—fails the EU originality threshold.

This EU perspective mirrors many of the concerns raised in the U.S. comments, suggesting a converging international view that the human‑in‑the‑loop model is the most defensible basis for protection. Practitioners operating across the Atlantic should therefore anticipate similar disclosure and authorship standards in forthcoming EU rulemaking.

6. Practical Implications for Creators and Registrants

Given the current U.S. guidance and the emerging EU analysis, creators who incorporate AI into their workflow should adopt a disciplined approach:

Document Human Decisions – Keep a contemporaneous log of prompt design, parameter selection, and post‑generation editing. This documentation can serve as evidence of human authorship if a registration is challenged.
Use the New Disclosure Fields – When filing a registration, complete the “AI‑Generated Content” checkbox and attach a concise description of the AI tool, version, and the nature of its contribution. This aligns with the Office’s interim policy and reduces the risk of a refusal.
Separate Human‑Created Elements – If possible, structure the final work so that the human contribution is clearly identifiable (e.g., a preface, commentary, or curated selection). This strengthens the claim that the work contains “a substantial amount of human‑created expression.”
Assess Derivative Risks – Before training an AI model on existing copyrighted material, verify that the source works are either in the public domain or covered by a license that permits such use.
Monitor International Developments – For works that will be distributed in the EU, prepare to meet the “personal intellectual contribution” test by documenting the creative choices that guided the AI output.

By treating the current guidance as a de‑facto standard, creators can avoid registration delays and position themselves for smoother compliance when formal rulemaking arrives.

7. Navigating the Rulemaking Process

The Copyright Office’s rulemaking follows the Administrative Procedure Act (APA). Stakeholders who wish to influence the final regulations should:

Submit Formal Comments – Comments must be filed within the comment period announced in the NOI (see [Artificial Intelligence and Copyright], 2024). Comments become part of the public record and are considered by the Office when drafting the final rule.
Provide Empirical Data – The NOI explicitly requests statistics on AI‑related registration requests. Supplying concrete numbers (e.g., “X % of registrations in 2023 included AI‑generated material”) strengthens a comment’s impact.
Engage in Public Hearings – The Office may hold webinars or public meetings; participation offers an additional avenue to ask questions and clarify positions.
Track Draft Regulations – After the comment period closes, the Office will publish a “Notice of Proposed Rulemaking” (NPRM). Stakeholders can again comment on the NPRM before a final rule is issued.

Because the rulemaking timeline can span several months, early and well‑documented participation is essential for shaping policy that reflects real‑world practice.

8. Checklist for AI‑Related Copyright Compliance

| ✅ Action | Why It Matters | How to Implement | |---|---|---| | Identify the human author(s) | Satisfies the statutory “author” requirement ([Copyright Registration Guidance], 2024) | Keep a record of who made the creative decisions (prompt design, editing). | | Complete the AI‑Generated Content disclosure | Aligns with the Office’s interim policy ([Copyright Registration Guidance], 2024) | Tick the new checkbox on the registration form; attach a brief description of the AI tool and its role. | | Document AI tool details (model, version, prompts) | Provides evidence of human direction and aids future audits | Save prompt logs, model version numbers, and any post‑processing steps in a dated file. | | Separate human‑created elements | Strengthens the “substantial human expression” test ([Comments In Response To The Copyright Office's Notice Of Inquiry On Artificial Intelligence