Mengchun’s Comments:
Several Chinese courts have recognized that AI-generated content may qualify for copyright protection, but only under two key conditions: (1) the creator must record and provide evidence of the creative process, and (2) the final output must demonstrate originality and individualized human input. This approach aligns with the fundamental principle of copyright law—protecting the intellectual contributions of natural persons.
Key Issue:
Under what circumstances can content generated by AI drawing software be protected as a work under Chinese copyright law?
Case Overview:
On August 15, 2023, a user named Feng published a series of AI-generated images titled “Phantom Wings Transparent Art Chair” on the Xiaohongshu platform, stating that they were created using Midjourney and sharing the English prompts used.
Five days later, Zhu contacted Feng expressing interest in mass-producing the designs, but no agreement was reached.
On August 22, 2023, Zhu subscribed to Midjourney and generated 40 product design drafts. After several rounds of refinement with a third-party design company, Zhu finalized a product design and produced a 3D-printed version with slight adjustments.
On January 19, 2024, Zhu published photos of the final product—referred to as the “Butterfly Chair”—on Xiaohongshu and linked to a sales page on his company’s store.
Feng later filed a lawsuit against Zhu and the associated company, claiming that his published images were original artworks protected by copyright and that the butterfly chairs, marketing images, and packaging designs constituted substantial similarity and thus copyright infringement and unfair competition.
Court Findings:
The core issue was whether Feng’s three published images met the threshold for protection as “works” under copyright law—specifically, whether they reflected the intellectual creativity of a natural person or were primarily generated by AI.
In general, copyright protects expressions that are independently created and reflect the author’s personal choices and creativity. When using text-to-image software like Midjourney, users can generate draft images simply by inputting basic prompts, with the AI playing a key role in transforming text into visual output. Without multiple iterations or deeper human involvement, such images often lack sufficient personal and creative input.
Feng provided no original documentation of the creative process, such as prompt history, workflow records, or evidence of parameter adjustments. Although he demonstrated a similar image generation process on August 5, 2024, the demonstration was not sufficient to prove a consistent and reproducible creative path leading to the disputed images.
The court emphasized that to establish authorship over AI-generated content, the user must show aesthetic choices and individualized contributions—for instance, by modifying prompts, selecting among results, adjusting parameters, or editing compositional elements such as layout, color, perspective, and line work. Absent such evidence, the images cannot be deemed to embody a natural person’s intellectual creation.
Decision:
The court found that:
- Feng’s three butterfly chair images lacked sufficient evidence of originality and individualized intellectual input.
- Without a traceable and verifiable creative process, the images did not meet the legal definition of “works” under Chinese copyright law.
- The court therefore dismissed all of Feng’s claims, including copyright infringement and unfair competition.
Takeaway:
Chinese courts are increasingly open to the protection of AI-assisted creations, but the burden remains on the user to demonstrate substantial human involvement. Notarized workflow records, creative logs, and documented iterations are essential to show that the final product reflects a human author’s original and individualized expression.
Court:
Zhangjiagang People’s Court
Final Decision Date:
March 19, 2025
Case No.:
（2024）苏0582民初9015号
(2024) Su 0582 Min Chu No. 9015
