On January 29, 2025, the U.S. Copyright Office released Part Two of its planned reports on the intersection between copyright and generative Artificial Intelligence (“AI”). Titled “Part 2: Copyrightability,” the Report addresses a hot issue—the extent to which AI-generated outputs are entitled to copyright protection. The Report, after examining the existing legal framework—under which only material generated by humans is copyrightable—concludes that the law does not need to change. Part One, discussed previously on this blog, focused on digital replicas. Both publications follow a lengthy notice and comment period. As the Report notes, “approximately half” of the public comments submitted to the Copyright Office “addressed copyrightability.”[1]
Copyright Protection Is For Humans
The Report opens with the Copyright Office’s observation that it has wrestled with difficult questions of authorship—a key requisite element for copyright protection—in the face of technological developments since as “early as 1965.”[2] The Copyright Office outlines two “bright-line rules” for approaching the question as it relates to generative AI:
Rule #1 — Work created by a human author with AI assistance can be protected by copyright: “Where AI merely assists an author in the creative process, its use does not change the copyrightability of the output.”[3]
Rule #2 — Work created entirely by AI cannot be protected by copyright: “[I]f content is entirely generated by AI, it cannot be protected by copyright.”[4]
The Report outlines the existing legal framework for the copyrightability of AI-generated works, including an overview of relevant case law, quoting, with emphasis added, the standard established by the U.S. Supreme Court in 1989 in a case entitled Community for Creative Non-Violence v. Reed: “the author [of a copyrighted work] is . . . the person who translates an idea into a fixed, tangible expression entitled to copyright protection.”[5] The Copyright Office observes that, to date, “[n]o court has recognized copyright in material created by non-humans, and those that have spoken on the issue have rejected the possibility.”[6]
Generative AI: Potential Uses and Copyright Implications
In its original Notice of Inquiry issued to the public on August 30, 2023, soliciting comment on topics related to generative AI,[7] the Copyright Office defined an “AI system” as a “software product or service that substantially incorporates one or more AI models and is designed for use by an end-user.”[8] In describing existing generative AI technology, the Report explains that “the output of current generative AI systems may include content that was not specified and exclude content that was,” and notes that some commenters described generative AI as a “black box.”[9]
Against the backdrop of generative AI’s expansive possible outputs and the technology’s opaqueness to some, the Copyright Office examines generative AI’s potential uses and their copyright implications in situations that do not neatly fit either of the Copyright Office’s two bright-line scenarios of authorship (i.e., (1) human-authored works created with AI assistance (copyrightable) or (2) entirely AI-generated works (not copyrightable)), providing the following guidance in the Report:
- Assistive Uses of AI Systems — “Computer-assisted tools,” such as those used in post-production (e.g., to age or de-age an actor), for visual effects, or Autotune in the music industry, are common.[10] Copyright protection depends on how the AI system is used, and the most important factor in that equation is the “distinction between using AI as a tool to assist in the creation of works and using AI as a stand-in for human creativity.”[11] Material created by a person may be copyrightable even if the creator used AI as a “brainstorming tool,” such as, for example, for song ideation or to create a preliminary outline.[12]
- Prompts to AI Systems — The Report “concludes that, given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output,” because “[p]rompts essentially function as instructions that convey unprotectible ideas” and users “do not control how the AI system processes them in generating the output.”[13] The Copyright Office itself experimented with generative AI technology, and notes that “identical prompts can generate multiple different outputs,” which “further indicates a lack of human control.”[14] However, the Report leaves the door open to a future in which works produced by prompts to AI systems may potentially be copyrightable, if those “prompts can sufficiently control expressive elements in AI-generated outputs to reflect human authorship.”[15]
- Expressive Inputs — An expressive input is one where a human author provides his or her own input (such as “text, images, audio, video, or a combination of those mediums”) and requests that AI modify that input, such as changing colors in a visual work or converting a first-person story to third-person.[16] The Report finds that the human author’s “own creative expression will be protected by copyright, with a scope analogous to that in a derivative work,” but the copyright cannot extend to the “AI-generated elements standing alone.”[17]
- Modifying or Arranging AI–Generated Content — “Generating content with AI is often an initial or intermediate step, and human authorship may be added in the final product.”[18] As in expressive inputs, the Report finds that “the copyright would extend to the material the human author contributed but would not extend to the underlying AI-generated content itself.”[19] The Report notes that this is often a “case-by-case determination,” but reassures creators that “the inclusion of elements of AI-generated content in a larger human-authored work does not affect the copyrightability of the larger human-authored work as a whole.”[20]
No Need For Change, For Now
The Report examines whether existing copyright law should change to address new AI technologies. While some commenters urged that interpretation of the Copyright Clause[21] “evolve,” the Report states that “[m]ost commenters that opined on this issue . . . agreed with the [Copyright] Office’s view that the Copyright Clause requires human authorship.”[22] The Copyright Clause refers to “promoting progress,” but commenters stated that AI systems “are inanimate objects that do not need an incentive to create” and “there appear to be sufficient incentives for AI companies under existing law.”[23] Commenters were concerned that granting AI works copyright protection would discourage human authorship.[24]
The Report examines a few specific concerns it received from commenters and their proposals:
- Sui Generis Rights — The Report rejects a proposal for new sui generis rights, finding the policy arguments unpersuasive.[25] Meaningful incentives already exist to develop AI models and systems, such as “patent, copyright, and trade-secret protection for the machinery and software, as well as potential funding and first-mover advantages” and additional incentives do not appear to be needed.[26] The Copyright Office also has concerns about “the impact of AI-generated materials on human authors and the value that their expression provides to society,” with AI-generated content flooding the market.[27]
- Creators with Disabilities — Some commenters argued that creators with physical and cognitive disabilities specifically benefit from the use of AI.[28] In response, the Copyright Office notes that “[c]opyright protection remains available where AI functions as an assistive tool that allows human authors to express their creativity.”[29]
- International Competition — Commenters raised concerns that the United States would fall behind in “international competition” if creators could not “exploit the economic value” of AI-generated works.[30] To that, the Copyright Office reminds us that “the United States is bound by our own Constitution and copyright principles.”[31] The Report also analyzes international views on copyright protections in AI-generated works, and observes that countries that “have addressed this issue so far have agreed that copyright requires human authorship.”[32] The Copyright Office intends to monitor developments abroad.[33]
The Report concludes by restating that the Copyright Office does not “recommend a change in the law.”[34] However, it intends to issue “additional registration guidance and update[e] the relevant sections of the Compendium of U.S. Copyright Office Practices.”[35]
* * * *
As discussed previously on this blog, the Copyright Office intends to release Part 3 of its Report, on the use of copyrighted works to train AI models, later in the first quarter of 2025. We will continue to monitor for the Copyright Office’s release of the remaining Parts of its Report.
[1] Report at ii.
[2] Report at 2.
[3] Report at 2.
[4] Report at 2.
[5] Report at 7.
[6] Report at 7.
[7] Specifically, (1) the use of copyrighted work to train generative AI models; (2) the copyrightability of materials generated using AI systems; (3) the potential liability for infringing AI-generated works; and (4) the treatment of generative AI outputs that imitate the identity or style of human artists. See 88 Fed. Reg. 59,942 at 59,945 (Aug. 30, 2023).
[8] Report at 5.
[9] Report at 6.
[10] Report at 11.
[11] Report at 12.
[12] Report at 12.
[13] Report at 18.
[14] Report at 20.
[15] Report at 21.
[16] Report at 22.
[17] Report at 24.
[18] Report at 24.
[19] Report at 25.
[20] Report at 27.
[21] “The Congress shall have Power . . . To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Investors the exclusive Right to their respective Writings and Discoveries.” U.S. Const. art 1, § 8, cl. 8.
[22] Report at 32.
[23] Report at 33-34 (quotations omitted).
[24] Report at 34-35.
[25] Report at 36.
[26] Report at 36.
[27] Report at 36.
[28] Report at 37-38.
[29] Report at 38.
[30] Report at 38-39 (quotations omitted).
[31] Report at 39.
[32] Report at 28.
[33] Report at 31.
[34] Report at 40.
[35] Report at 41.