Per Yuan Hao of Berkeley Law, the Beijing Internet Court recently released a translation of Li vs. Liu in which the Court recognized copyright in images created by generative AI. The plaintiff, Mr. Li, used Stable Diffusion (an artificial intelligence) to generate the image involved in the case and published it on the Xiaohongshu (Little Red Book) platform; the defendant, a blogger on Baijiahao, used the image generated by the plaintiff’s AI to accompany the article, and the plaintiff sued. The Court held that the artificial intelligence-generated image involved in the case met the requirements of “originality” and reflected a human’s original intellectual investment and should be recognized as works and protected by copyright law. This is the opposite of the decision reached by the U.S. Copyright Office in Zarya of the Dawn (Registration # VAu001480196) that did not recognize copyright in AI-generated images. Note this Beijing case is also different from the recent Thaler v. Perlmutter decision (Civil Action No. 22-1564 (BAH)) because Thaler was trying to recognize the AI as the author and not the person using the AI as a tool as author. Note that the Shenzhen Nanshan District People’s Court ruled similarly to the Beijing Internet Court for AI-generated news articles in 2020.

Excerpts from the translation follow. The full translation is available here.

On February 26, 2023, the plaintiff posted the picture involved (see Figure 6) on his Little Red Book account “****” ( Little Red Book ID: ******), and the title of the post was “Spring Breeze Brings Tenderness.” Apart from the picture involved, the post include d five other pictures,with hashtags #AI#[topic]##AIillustration#AIpainting#photo#girl#photography#spring#beauty. Through investigation, it is found that the plaintiff is able to log in to the account and check the picture involved by obtaining a verification code through his mobile phone. The plaintiff claimed that the picture was generated via Stable Diffusion on February 24, 2023. He submitted a video that demonstrates the process of generating the picture involved…

The plaintiff believes that the defendant used the picture involved without his permission and cut off his signature watermark on Little Red Book , which infringed his right of authorship and right of dissemination on the information network regarding the picture.

The Court reasoned,

1. Whether the picture “Spring Breeze Brought Tenderness” constitutes a work and what type of work it constitutes

According to Article 3 of the Copyright Law of the People’s Republic of China (hereinafter referred to as the Copyright Law ), “The works mentioned in this Law refer to intellectual achievements that are original and can be expressed in a certain form in the fields of literature, art, and science,” when examining whether the object for which the plaintiff claims copyright constitutes a work, the following elements should be considered: 1 . Whether it falls under the realm of literature, art, or science; 2. Whether it is original; 3. Whether it is expressed in a certain form; 4. Whether it is an intellectual achievement. In this case, the pictures involved is no different from the photos an d paintings that people usually see; obviously it falls under the category of art and is expressed in a certain form, so elements 1 and 3 are met.

“Intellectual achievements” refer to the results of intellectual activities, so the work should reflect the intellectual input of a natural person. In this case, the plaintiff used the hashtag “AI illustration” when publishing the picture involved; and the plaintiff could reproduce the process of generating the picture involved using the Stable Diffusion model and the prompt words and parameters set by himself. Unless there is contrary evidence, it can be found that the picture “Spring Breeze Brings Tenderness” is generated by the plaintiff using AI. According to public information and relevant research, the Stab le Diffusion model is trained from a large number of pictures and their corresponding text descriptions on the Internet. Based on the text instructions, the model can use the correspondence between the semantic information in the text and the pixels in the picture to generate a picture that matches the text. This picture is not a ready made one that can be obtained through a search engine, nor is it an arrangement or combination of various elements preset by the software designer. In layman’s terms, the Stable Diffusion model works in a way that a human does: it acquires some abilities and skills through learning and accumulation, and it can generate a picture based on the text descriptions input by humans drawing the lines and doing the colors, and presenting man’s creative ideas in a tangible way. In this case, the plaintiff wanted a close up of a beautiful woman under dusk light, so he entered the following prompt words into the Stable Diffusion model: “ultra photorealistic” and “color photo” for the art type; “Japan idol” for the subject, along with detailed description of the character such as skin, eyes, and braid color; “in locations”, “golden hour”, and “dynamic lighting” for the environment; “cool pose” and “viewing at camera” for the way the character is presented; and “film texture” and “film simulation” for the style. The parameters were also set. Based on the initially generated picture, the plaintiff added some prompt words, modified the parameters, and finally got the picture he wanted. From the time the plaintiff had an idea about the picture to his final selection of the picture involved, the plaintiff did some intellectual investment, such as designing the presentation of the character, selecting prompt words, arranging the order of prompt words, setting parameters, and selecting the picture that he wanted. The picture involved reflects the plaintiff’s intellectual investment, so it meets the element of “intellectual achievement”.

Of course, not all intellectual achievements are works; only those with “originality” are. Generally speaking, “originality” requires that the work be completed independently by the author and reflect the author’s personalized expression. “Mechanical intellectual achievements” are excluded. For example, if a work is completed based on a certain order,
formula, or structure, different people will get the same result; as the expression is singular, the work does not have originality. And one has to decide according to the specific situation whether an AI generated picture reflects the author’s personalized expression. Generally speaking, when people use the Stable Diffusion model to generate pictures, the more different their needs are and the more specific the description of picture elements, layout, and composition is , the more personalized the picture will become. In this case, there are identifiable differences between the picture involved and the prior works. In terms of the generation process of the picture involved, the plaintiff did not draw the lines himself, or instruct the Stable Diffusion model everything on how to draw the lines and do the colors; the lines and colors that constitute the picture involved are basically done by the Stable Diffusion model, which is very different from the conventional way of people using brushes or software to draw pictures. However, the plaintiff used prompt words to work on the picture elements such as the character and how to present it, and set parameters to work on the picture layout and composition, which reflects the plaintiff’s choice and arrangement. The plaintiff input prompt words and set parameters and got the first picture; then he added some prompt words, modified the parameters, and finally got the picture involved. Such adjustment and modification also reflect the plaintiff’s aesthetic choice and personal judgment. During the trial, the plaintiff generated different pictures by changing the prompt words or the parameters. One can infer that with this model, different people can generate different pictures by entering different prompt words and setting different parameters. Therefore, the picture involved is not a “mechanical intellectual achievement”. Unless there is contrary evidence, it can be found that the picture involved is independently completed by the plaintiff and reflects the plaintiff’s personalized expression. In summary, the picture involved meets the element of “originality”.

A new generation of generative AI technology is being used by more people for creation. The Stable Diffusion model and models with similar functions can generate beautiful pictures based on text descriptions. Many people, including those without drawing skills, are trying to use these new models to present their creativity and designs in a tangible way; and the models have greatly imp roved the efficiency of picture creation. The generative AI technology has changed the way people create. Just like many other technological advances in history, the process of technological development is the process of outsourcing human work to machines. Before the advent of cameras, people needed superb painting skills to reproduce an object perfectly; then the cameras made it easier to record the image of an object. Nowadays, the camera of smartphones is getting better and easier to use. However, as long as the photos taken with a smartphone reflect the photographer’s original intellectual investment, they will constitute photographic works and are protected by the Copyright Law. The development of technologies and tools require less human investment, but the copyright system should remain in use in order to encourage the creation of works. Before the emergence of the AI model involved, people needed to spend time and energy learning how to paint, or to consign others to paint for them. In the second scenario, the painter will draw the lines and fill in the colors upon the client’s request to complete a work of fine art. And the person who draws is normally considered a creator. This is similar to the use of AI models to generate pictures, but there is one major difference here: the creator has his own will and he will use some judgment when painting for the client. Currently, the generative AI model has no free will and is not a legal subject. Therefore, when people use an AI model to generate pictures, there is no question about who is the creator. In essence, it is a process of man using tools to create, that is, it is man who does intellectual investment throughout the creation process, the not AI model. The core purpose of the copyright system is to encourage creation. And creation and AI technology can only prosper by properly applying the copyright system and using the legal means to encourage more people to use the latest tools to create. Under such context, as long as the AI generated images can reflect people’s original intellectual investment, they should be recognized as works and protected by the Copyright Law.

To sum up, the picture involved meets the definition of a work and should be considered as such. As to what type of work it constitutes, the plaintiff claims that it is a work of art; and if the Court does not think so, then it should be seen as “other intellectual achievements that have the characteristics of a work”. In judicial practice, when judging the type of a work, the first thing i s to determine whether it falls under the types of work listed in the Copyright Law. Specifically, the Court should compare the characteristics and expression of the work involved with those listed in the first eight items of Article 3 of the Copyright Law. If the work falls under any of the types of works listed in the first eight items, then the Court will
identify it as that type of work; and the ninth item “other intellectual achievements consistent with the characteristics of a work” will no longer apply. According to Article 4 of the Regulations for the Implementation of the Copyright Law of the People’s Republic of China: “Fine art refers to paintings, calligraphy, sculptures and other aesthetically significant two dimensional or three dimensional works composed of lines, colors, or other methods.” In this case, the picture involved is a graphic art work that is composed of lines and colors and is of aesthetic significance, so it is fine art. As there is no need to apply the “other works clause” to protect the picture involved, it is not “other intellectual achievements that have the
characteristics of a work”. To sum up, the picture involved is fine art and shall be protected by the Copyright Law.

2. Whether the plaintiff owns the copyright on the picture involved
Paragraph 1 of Article 11 of the Copyright Law stipulates that: “Copyright belongs to the author, unless otherwise provided for in this Law.” Article 11 of the Copyright Law stipulates that: “An author is a natural person who creates a work. The author is the natural person who creates the work. For works hosted by a corporate or unincorporated organization, created on
behalf of the will of the corporate or unincorporated organization, and for which the corporate or unincorporated organization assumes responsibility, the corporate or unincorporated organization shall be regarded as the author.” It suggests that an author can only be a natural person, a corporate or unincorporated organization; that is consistent with the civil subjects stipulate d in the Civil Code. Therefore, an artificial intelligence model cannot be deemed as an author under China’s copyright law. As a result, although the picture involved is “drawn” by the artificial intelligence model involved, the model is not the author of the picture.
The designer of the artificial intelligence model involved neither had the intention to create the picture involved, nor did he preset the content generated afterwards. He did not involve in the generation process of the pictures involved; so in this case, he is only a producer of the creation tool. By designing the algorithm and model and using a large amount of data to “train” it, the designer has equipped the AI model with the ability to autonomously generate content in response to different needs. The designer has undoubtedly done some intellectual investment during that process, but such investment has gone to the design of the AI model, that is, the production of a “creation tool”, not the picture involved. Therefore, the designer of the AI model involved is not the author of the picture involved.

In addition, according to the evidence in record, the designer of the AI model involved states in the license it provided that it “does not claim rights to the output [of the model].” It can be determined that the designer claims no right in relation to the output of the model.

As mentioned above, the plaintiff is the one who directly set up the AI model involved as needed and finally selected the picture involved. The picture involved is generated directly due to the plaintiff’s intellectual investment and it reflects the plaintiff’s personalized expression. Therefore, the plaintiff is the author of the picture involved and owns the copyright on it.

It should be noted that although the Court finds that the plaintiff, as the author, owns the copyright on the picture involved, the plaintiff should prominently mark the AI technology or model used in line the principle of good faith and the need to protect the public’s right to know. In this case, the plaintiff uses the hashtag “AI illustration”, which is enough to let the public know that the content is generated by the plaintiff using AI technology. The Court recognizes this to be a proper practice.

…