On 27 November 2023, the Beijing Internet Court delivered a first-instance judgment on the copyright infringement dispute of AI-generated images, which is the first case of copyright in the field related to AI-generated images.
Case History
On 25 May 2023, the Plaintiff, Mr. Li sued the Defendant, Ms. Liu, against Ms. Liu’s infringement on his right to authorship of his works and his right to disseminate information on the Internet. On 24 August 2023, the Beijing Internet Court (Court) heard the case publicly under a full-media live broadcast.
The Plaintiff claimed that the Defendant used in the article, without his permission, an image (Image at issue) that he generated and published by typing in prompt words using the open source software Stable Diffusion, and accordingly requested the Court to order the Defendant to apologize to him, eliminate the impact of the infringement on him, and compensate him for the economic loss of RMB5,000.
The Defendant argued that she obtained the Image through an internet search and was unsure whether the Plaintiff had rights to this Image. The main content of her posting was original poems, not the Image, and there was no commercial use, so she had no intent to infringe. In addition, the Defendant also argued that the market price of AI-generated images was very low, while the amount of economic compensation claimed by the Plaintiff was too high.
The First-instance Decision
The Court found the following three points of contention in this case and commented on each of them as follows:
i. Whether the Image at issue constitutes a work and what type of work it constitutes
Article 3 of the PRC Copyright Law stipulates that “the works referred to in this Law refer to intellectual achievements in the fields of literature, art and science that are original and capable of being expressed in a certain form.”
As to “intellectual achievements”, the Court held that the Image at issue was generated by the Plaintiff using generative artificial intelligence techniques. From the time the Plaintiff conceived the Image to the final selection of the Image, the Plaintiff made certain intellectual investments, such as designing the presentation of the characters, selecting the cues, arranging the order of the cues, setting the relevant parameters, and selecting which images would meet expectations. Thus, the Image that reflected the Plaintiff’s intellectual input should be regarded as his “intellectual achievements”.
As to “originality”, the Court held that the Image at issue was not a “mechanical intellectual product”, as the model was used to create different content by different people entering different prompts and setting different parameters. Therefore, it could be concluded that the Images was independently created by the Plaintiff, and was “original”.
Article 4 of the Implementing Regulations of the Copyright Law stipulates that “fine arts works refer to paintings, calligraphy, sculptures and other works of plastic arts in plane or in three dimensions that have aesthetic significance and are composed of lines, colors or in other ways.” In this case, the Image at issue is a graphic plastic art work of aesthetic significance composed of lines and colors, which is concluded by the Court as a work of fine art.
ii. Whether the Plaintiff has a copyright in the Image at issue
According to Article 11 of the Copyright Law, authors are limited to natural persons, legal entities or organizations. The AI model itself cannot be an author under Chinese copyright law. Because of this, although the Image at issue was “drawn” by the AI model, the model could not be the author of the Image.
And the AI model designer was not the author of the Image either, considering 1) the designer had no desire to create the image, nor predetermine the subsequent generation of content, but was merely the producer of the creation tool, 2) the designer of the AI model in this case stated in the license that it “does not claim rights to the output”.
As mentioned above, the Plaintiff was the one who directly made the relevant settings to the AI model as needed and ultimately selected the Image at issue, which was directly generated based on Plaintiff’s intellectual input and reflect his personalized expression. Therefore, the court found that the Plaintiff was the author of the Image at issue and enjoyed the copyright of the Image.
iii. Whether the Defendant’s acts constitute infringement and whether the Defendant should be held liable
In this case, the Defendant, without authorization, used the Image at issue as an accompanying picture and posted it in its own account, which infringed the Plaintiff’s right to disseminate information network in respect of the Image. And the Defendant’s removal of the watermark violated Plaintiff’s right of attribution.
Based on the above, the Plaintiff’s request to order the Defendant to publish a statement apologizing to the Plaintiff, and eliminate the effects of its infringement on the Plaintiff, on the platform where the Defendant used the Image, was supported by the Court, since this was comparable to the scope of the impact of the Defendant’s infringement on the Plaintiff.
The court finally determined that the Defendant should compensate the Plaintiff for the economic loss in the amount of RMB500 for the infringement, based on the situations of the Image involved and the infringing use.
The Court’s Additional Clarifications
Due to the high profile of this case, the Court also clarified the following important matters in its judgment:
i. The essence of people using AI models to generate images is that people use tools to create images. In other words, it is the person who makes the intellectual investment in the entire creation process, not the AI model. Images generated by AI should be recognized as works and protected by copyright law, as long as they reflect the original intellectual input of the person.
ii. The designers of AI models design algorithms and models, and use a large amount of data to “train” the AI, so that the AI model has the ability to generate content on its own in the face of different needs, and in this process, there must be an intellectual investment, but the designer’s intellectual investment is embodied in the design of the AI model, which is embodied in the production of the “creative tools”, rather than in the generated pictures.
In accordance with the principle of good faith and the need to protect the public’s right to know, users of AI models should prominently label the AI technology or model they are using, so as to let the public know that the content is generated using AI technology.
By MMLC Group, China
Law Firm Website: https://www.mmlcgroup.com
Source:
https://www.hg.org/legal-articles/china-s-first-ai-generated-image-copyright-infringement-case-65918