China - Li v Liu, 27 November 2023

Claimant: Li

Defendant: Liu

The claimant sought the following remedies:

1. Remove infringing videos;
2. Prominently display a publicly issued apology statement within the applet for one consecutive month;
3. Compensate the plaintiff for economic losses.

Claimant:

1. The image in question is copyrighted work

• Model selection and acquisition: The claimant independently selected the AI model based on aesthetic preferences.
• Prompt input:  The claimant crafted and entered specific prompts, including reverse prompts to exclude certain elements, reflecting their creative judgement and design.
• Generation parameters: The claimant determined parameters such as sampling methods and guidance coefficients through repeated experimentation, demonstrating intellectual effort and originality.

Based on the above, the claimant claimed that image in question therefore constituted an original intellectual creation and is therefore protected under copyright law.

2. The defendant infringed the claimant’s right of authorship and information network transmission right

The defendant, without authorisation, removed the claimant’s attribution watermark and disseminated the image, misleading users into believing the defendant was the author. This infringed the claimant’s rights of authorship and dissemination via information networks.

Defendant:

The defendant merely argued that AI-generated images have low market value and did not present any substantive defence.

Court’s opinions:

1. AI-generated works should be recognised as copyrighted works and protected under copyright law, provided they demonstrate original intellectual input by a human.

AI-generated works may be protected under copyright law if they reflect original human intellectual input. Unlike commissioned artists, AI lacks free will and legal subjectivity. Therefore, the creative input must be attributed to the human configuring and directing the AI tool.

2. The images in question constitute works of fine art and are protected under copyright law.

1. Intellectual achievement
   The image was not sourced from existing stock or pre-set templates. The claimant’s involvement—from conceptualisation to prompt design and parameter setting—demonstrates intellectual input.
2. Originality
   The claimant refined the image through iterative prompt and parameter adjustments, reflecting personal aesthetic judgement. The process differs from mechanical or formulaic outputs, and the final image embodies the claimant’s unique expression.

3. The plaintiff holds the copyright but should prominently label the AI technology or model employed

The claimant, having directly configured the AI model and selected the final output, is the author and copyright holder of the image. However, in line with the principle of good faith and the public’s right to know, the claimant is obliged to clearly indicate the AI technology or model used in the creation process.

Judgement:

The defendant shall publish an apology to the plaintiff on the social media account in question, maintaining it for no less than 24 hours to mitigate the impact, and shall compensate the plaintiff for economic losses.