Noted with Interest: Copyrightability of AI-Generated Content in China and the US

Since November 2023, there have been several important decisions in the Chinese court system granting AI users copyrights to AI-generated content. In contrast, the U.S. Copyright Office (“USCO”) has historically hesitated to grant copyrights to AI-generated images, pointing to a lack of human authorship and contribution. Then, on January 29, 2025, the USCO released a “Report on Copyright and Artificial Intelligence,” which emphasized that, where “creativity is expressed through the use of AI systems, it continues to enjoy protection.” Just one day later, on January 30, USCO granted for the first time copyright protection to an AI image titled “A Single Piece of American Cheese.” The USCO’s release of the “Report on Copyright and Artificial Intelligence” and subsequent grant of copyright protection to an AI image demonstrates a convergence of Chinese and U.S. law that certain AI-generated content warrants copyright protection though differences in the applicable analysis remain.

The Chinese Court Decisions

The Beijing Internet Court Case—On November 27, 2023, the Beijing Internet Court (BIC) issued a decision recognizing copyright in AI-generated images. There, the plaintiff generated the images at issue using an open-source generative AI model called Stable Diffusion with textual prompts and published the images on a first platform (Xiaohongshu). The defendant later used one of the AI-generated images in her article on a different, second platform (Baijiahao) without plaintiff’s permission.

The BIC decided that the author put in a sufficient amount of intellectual investment, such as selecting over 150 prompts, arranging their order and setting specific parameters, and continuing to adjust and modify prompts and parameters until the final image aligned with his conception. The BIC found that the resultant images were thus not just “mechanical intellectual achievements” but rather were independently completed by the author and reflected the author’s personalized expression, thereby qualifying for copyright protection.

The Zhangjiajie Case—In contrast, on March 19, 2025, the Zhangjiagang People’s Court issued a decision dismissing a copyright infringement and unfair competition suit brought by a designer against furniture manufacturers and distributors. There, the plaintiff created three images using the AI image generation tool Midjourney. These images depicted a butterfly-shaped chair with a jelly-like texture in varying hues of pink, bluey and orange. Seeking to attract interest from furniture manufacturers, the plaintiff posted the images on social media along with the prompts used to generate them. Defendants in the case approached the plaintiff expressing interest in the design, but the parties failed to reach an agreement. Despite this, the defendants proceeded to use Midjourney to recreate similar images using plaintiff’s publicly shared prompts before manufacturing and selling butterfly chairs based on a final design.

The court ruled in favor of the defendants on the grounds that the plaintiff’s AI-generated images were not protected by copyright and that the defendants were not liable for copyright infringement. The court differentiated content produced by a user exercising their “original human intellectual” input in the creation process with the assistance of AI as a tool with content mainly autogenerated by AI with minimal human contribution. Only the former is protectable by copyright.

The USCO Guidance

The 2023 USCO Registration Guidance

In March 2023, the USCO issued a guidance on works containing material generated by AI, taking the position that “authorship” requirement of US copyright law refers to “human authorship,” and that if a work contains more than a de minimis amount of AI-generated material, the applicant must disclose that information and provide a brief statement describing the human author’s contribution.

Copyright Decisions After the 2023 USCO Registration Guidance

USCO copyright registration decisions following the 2023 USCO registration guidance routinely denied copyright registration to material generated using AI unless the human user subsequently made further, substantial changes without AI.

A Recent Entrance to Paradise (Feb. 2023) (Thaler v. Perlmutter, D.C. Circuit, Aug. 2023)—rejected work that lacked “traditional human authorship” and was “autonomously generated by an AI” as both copyrightability and registrability require human authorship.

 Zarya of the Dawn (Feb. 2023)—rejected images in a book that were generated by Midjourney, identifying “significant distance” between the user’s input and Midjourney’s output.

Théâtre D’opéra Spatial (Sept. 2023)—refused to register artwork partially generated by AI as the work contains more than a de minimis amount of AI-generated content that applicant refused to disclaim.

Suryast (Dec. 2023)—refused to register artwork generated by GenAI painting tool which was prompted to combine Vincent Van Gogh’s Starry Night with a photograph of a sunset because “the expressive elements of pictorial authorship were not provided by [the author].”

Rose Enigma (Mar. 2023)—granted partial protection to elements created directly by the artist by hand, while excluding portions of the work generated by AI on the grounds that they did not result from an autonomous human creative effort.

A Single Piece of American Cheese (Jan. 2025)—granted copyright registration to the digital graphic composition produced by AI with “inpainting” technique that allows artists manually to select and substantially modify parts of the work, resulting in human intervention sufficient to satisfy the creativity criteria required for copyright recognition.

2025 USCO Report: Copyright and Artificial Intelligence - Part 2: Copyrightability

USCO released on January 29, 2025 its report on the copyrightability of outputs generated by AI systems.  The 2025 Report emphasizes the necessity of human creativity in the creative process for copyrightability, addressing different scenarios of human contributions to AI-generated content. Assistive uses—The 2025 Report acknowledges the important distinction between “using AI as a tool to assist in the creation of works” and “using AI as a stand-in for human creativity,” stating that assistive uses that enhance human expression do not limit copyright protection. Using prompts—The 2025 Report highlights that merely using prompts to generate content from an AI model, even with highly detailed prompts or repeatedly revised prompts, “do not provide sufficient human control to make users of an AI system the authors of the output.” Even though “prompts may reflect a user’s mental conception or idea,” the 2025 Report decides that “they do not control the way that idea is expressed.” Using “expressive inputs”—When using creative work itself (such as original illustration, story, or other traditional copyrighted work) as inputs into an AI system, if that input work remains perceptible in the AI-generated output, the human can claim authorship of, and copyright in, the perceptible portion of the input work in the output. The 2025 Report noted that such expressive inputs, although they “may be seen as a form of prompts, are different from those that merely communicate desired outcomes.” Modifying or Arranging AI-Generated Content—The 2025 Report acknowledges human authorship may be added in the final product when AI is used as an initial or intermediate step, to the extent that “a human may select or arrange AI-generated material in a sufficiently creative way that ‘the resulting work as a whole constitutes an original work of authorship.’”

The 2025 USCO Report’s recognition that a human author may use AI as just another tool for generating a creative work entitled to copyright protection brings the USCO’s position more in line with the approach taken by Chinese courts. Indeed, in the wake of the 2025 USCO Report, the USCO granted for the first time copyright protection to an AI-generated work on January 30, 2025 – more than a year after Chinese courts first recognized copyright protection for AI images. However, even as courts and agencies on both sides of the Pacific begin to align on the basic concept that some AI-generated content qualifies for copyright protection, questions remain as to how much input is required to transition from content created predominantly by AI and thus not subject to copyright protection to content with a human author that is merely using AI as a tool. For example, Chinese courts, including the November 27, 2023 BIC case, have found that repeatedly adjusting and modifying prompts and parameters may lead to copyright for the resultant AI-generated image. In contrast, in the 2025 USCO Report, the USCO still “conclude[d] 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.” 2025 USCO Report at 18.