News and developments
How Copyright Responds to AI-generated Images
This case is noted to be the first of its kind in China, involving a plaintiff surnamed Li and a defendant surnamed Liu. Li claimed that Liu used an image of a woman in her personal webpage without his permission in February 2023. Li mentioned that he had generated the image using Stable Diffusion, an open-source software and posted it on his own social media account at Xiaohongshu, a social media platform widely regarded as the Chinese version of Instagram.
Liu argued that she found the image by using a search engine and did not see any watermark or information about its copyright owner. She also said she was not using the image for commercial purposes. However, the Beijing Internet Court eventually ruled in favor of Li, pointing out that the image showed “intellectual achievements” by Li. The court said that Li had input keywords on the expectations of the image, such as the appearance and posture of the young woman, filter of the image, and even the lighting. Li had also invested in adjusting the draft images to get the ultimate image that he had chosen.
*Note: The above figure as a whole shows the operating interface of Stable Diffusion, and the image embedded displays the output generated.
2.Key analysis: AI-generated image regarded as work
According to the ruling and like other copyright infringement disputes, the main issues typically addressed include:
(i) whether the picture in question constitutes a work under the Copyright Law of the People’s Republic of China and what type of work it constitutes;
(ii) whether the plaintiff owns the copyright on the disputed picture; and
(iii) whether the defendant’s behavior shall constitute an infringement. Compared to traditional copyright cases, particularly the ones occurring prior to the advent of artificial intelligence (the “AI”) and Large Language Model (the “LLM”), two terms “intellectual achievements” and “originality” are intensively discussed in the ruling of the current case.
According to the ruling, the court defined the term “intellectual achievements” as the results of intellectual activities, which reflects the intellectual input of a natural person. Compared to the operating model and logic of the LLM at issue, the court emphasized the “intellectual investment” the plaintiff made in the disputed picture, demonstrating that designing the presentation of the character, selecting prompt words, arranging the order of prompt words, setting parameters, and selecting the intended picture shall all be regarded as the “intellectual investment”.
With respect to the term “originality”, the court stated that “to meet the ‘originality’ requirement, the work shall be completed independently by the creator, reflecting such creator’s personalized expression.”[3] Historically speaking and even in the global legal landscape, the standard of “originality” is relatively low. In copyright cases, when pondering such subtle issue, it is commonly understood that the first of the two requirements encompassed by the term “originality” is independent creation and creativity. In the present case, the court claimed that “if (i) a work is completed based on a certain order, formula, or structure and (ii) different people would yield same results, the work does not have originality since the expression at issue is the only effective way of conveying a particular idea”.[4] Similarly, such principle already underlies the merger doctrine in the context of the western copyright system. According to the ruling of the present case, however, “when people use the Stable Diffusion model to generate images, diverse needs and expectations will be reflected in the picture once they specify and describe details of elements, layout, and compositions of the image”.[5] In this case, as the court finally determined that “there are identifiable distinctions between the picture in question and prior works”.[6] As a result, the court concluded that the work in question satisfied the requirement of originality.
It is also worth noting that the court forthrightly tied its ruling to a general policy argument, highlighting that the primary purpose of the copyright system is to encourage creation. Only if the copyright regime and legal methods could be properly applied, simulating more people to use the most up-to-date tools for creation process, more work will be cultivated, and the development of the AI technology will be enhanced. In addition, such policy analysis was heavily examined in the Research on Legal Feature and Copyright Attribution of AI-generated Images, where the presiding judge of the First AI-image Case continued to stress the competing considerations like the creator’s rights and interests as well as the industrial development shall be balanced when making the judgement.[7]
In this case, the USCO refused to register the Work because it “lack[ed] the human authorship necessary to support a copyright claim.”[9] The USCO examined the application by demonstrating the following guidance: “[Whether the ‘work’ is basically one of human authorship, with the computer [or other device] merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by a machine.”[10]
The USCO further argued that all of the Work’s “traditional elements of authorship” are generated by AI. As a result, the work lacks human authorship. If, however, a work containing AI-generated material also contains sufficient human authorship to support a claim to copyright.
Author: Xiangxiang MA
Footnotes [1] AIGC is the abbreviation of Artificial Intelligence Generated Content. [2] See the Civil Judgement of the Beijing Internet Court, file no.: (2023) Jing 0491 Min Chu No. 11279 (2023), https://english.bjinternetcourt.gov.cn/pdf/BeijingInternetCourtCivilJudgment112792023.pdf [3] See supra note 2. [4] See supra note 2. [5] See supra note 2. [6] See supra note 2. [7] ZHU Ge, The Research on Legal Feature and Copyright Attribution of AI-generated Images, Intellectual Property, 1 (2024) [8] See Re: Second Request for Reconsideration for Refusal to Register SURYAST (SR # 1-11016599571; Correspondence ID: 1-5PR2XKJ), https://www.copyright.gov/rulings-filings/review-board/docs/SURYAST.pdf. [9] See supra note 5. [10] See supra note 5. [11] See supra note 4.
![](https://www.legal500.com/developments/wp-content/uploads/sites/19/2024/07/Picture1-1.png)
- Different angles: when “Originality” interpreted in another jurisdiction
![](https://www.legal500.com/developments/wp-content/uploads/sites/19/2024/07/Picture2-1.png)
- Concluding observations
Author: Xiangxiang MA
Footnotes [1] AIGC is the abbreviation of Artificial Intelligence Generated Content. [2] See the Civil Judgement of the Beijing Internet Court, file no.: (2023) Jing 0491 Min Chu No. 11279 (2023), https://english.bjinternetcourt.gov.cn/pdf/BeijingInternetCourtCivilJudgment112792023.pdf [3] See supra note 2. [4] See supra note 2. [5] See supra note 2. [6] See supra note 2. [7] ZHU Ge, The Research on Legal Feature and Copyright Attribution of AI-generated Images, Intellectual Property, 1 (2024) [8] See Re: Second Request for Reconsideration for Refusal to Register SURYAST (SR # 1-11016599571; Correspondence ID: 1-5PR2XKJ), https://www.copyright.gov/rulings-filings/review-board/docs/SURYAST.pdf. [9] See supra note 5. [10] See supra note 5. [11] See supra note 4.