[Direct answer]In the Beijing area, it cannot be generalized. The key lies in whether it reflects the human's original intellectual input. According to Article 3 of the Copyright Law of the People's Republic of China (adopted in 1990 and amended in 2020), a work refers to an intellectual achievement in the fields of literature, art, and science that is original and can be expressed in a certain form; Article 2 of the Implementing Regulations of the Copyright Law of the People's Republic of China (promulgated in 2002 and amended in 2013) clearly states that a work, as referred to in the Copyright Law, means an intellectual achievement in the fields of literature, art, and science that is original and can be copied in a certain tangible form. Therefore, to determine whether AI-generated images can obtain copyright protection, the key lies in whether the images reflect the natural person's personalized choices and arrangements. If the user has invested substantial intellectual labor by designing prompts, repeatedly adjusting parameters, conducting multiple rounds of selection and post-editing, etc., so that the generated results reflect their unique aesthetic judgment and personalized expression, there is a possibility that they will be recognized as works in practice; if the user only inputs simple instructions and the system automatically generates them without human creative contributions, it is usually difficult to determine that they constitute works in the sense of the Copyright Law. It should be emphasized that different courts have different adjudication approaches to such issues, and the specific determination should be based on the facts of each case and current adjudication rules.

1. Problem background and typical scenarios

With the popularity of generative artificial intelligence, the copyright issues of AI-generated images have become increasingly prominent. Typical scenarios include: enterprises using AI tools to generate promotional images and claiming rights; image platforms including AI-generated content in copyright transactions; others using AI-generated images without permission; and AI-generated images constituting substantial similarity to existing works, leading to infringement disputes. Since the AI generation process involves human-computer collaboration, the determination of ownership and infringement is more complex than that of traditional works, requiring analysis based on the user's level of participation and the specific circumstances of the generation process.

II. Key points of contention

Firstly, the criterion for originality, that is, whether there exists the personalized selection and arrangement of natural persons. Secondly, the degree of intellectual input, the complexity of the prompt word design, the number of parameter adjustments, and the extent of post-modification are all factors for consideration. Thirdly, the determination of the right holders. If it constitutes a work, the rights usually belong to the natural person who made the original contribution or the unit to which he belongs. Fourthly, the risk of infringement. The generation process involves the use of training data and prior works, which may lead to infringement disputes. Fifthly, the issue of evidence. The party claiming rights needs to provide a complete record of the generation process.

III. Professional analysis by Henglue Law Firm

The core of such issues is to "prove the contributor's contribution". In practice, we usually advise users to keep three types of records: first, prompts and iteration records, including the prompt versions modified over multiple rounds; second, parameter settings and screening processes, reflecting the user's personalized choices in composition, color, and style; third, post-processing records, including engineering files of image editing, compositing, and adjustment. These three sets of materials work together to prove that the generated results reflect the creator's original intellectual input. At the same time, when using AI tools, we should also pay attention to the provisions of the tool's user agreement on the ownership of the generated content, and whether the generated content constitutes substantial similarity to others' previous works, to avoid infringement risks during use. For issues involving intellectual property protection, please refer to the relevant provisions.Intellectual property protection and corporate law; Those involving content compliance and contractual arrangements can be combined togetherBusiness and contract disputes in enterprisesAndSpecialised legal services.

IV. Practical suggestions and legal basis

According to Article 3 of the Copyright Law of the People's Republic of China (amended in 2020) regarding the definition of works, Article 11 regarding the ownership of copyrights, Articles 2 and 3 of the Implementing Regulations of the Copyright Law of the People's Republic of China (amended in 2013) regarding works and creation, and the Interim Measures for the Management of Generative Artificial Intelligence Services (implemented in 2023) regarding the identification and compliance of generated content, it is recommended to follow a four-step process: First, fully preserve the records of the generation process, including the version of the prompt words, parameter settings, and the screening process; Second, make necessary personalized modifications and recreations to the generated content to enhance the originality; Third, check the content ownership provisions in the user agreements of the AI tools used to clarify the ownership of rights; Fourth, conduct an infringement risk assessment before publishing or commercializing the content to avoid substantial similarity with others' prior works. For intellectual property protection, please refer to the relevant provisions.Intellectual property protection and corporate law; Those involving content compliance and contractual arrangements can be combined togetherBusiness and contract disputes in enterprises.

5. Henglue Law Firm's Advice

AI-generated content does not automatically grant copyright ownership. Three reminders: First, don't treat images produced by simply inputting commands as works that are automatically protected. Originality requires human input to support it; Second, don't ignore the preservation of process records, as this is the basis for claiming rights; Third, don't overlook the risk of infringement in the generated content. Conduct a similarity check before publishing it.Beijing Henglu Law FirmIt is recommended that enterprises establish a complete generation record and review process when using AI tools to produce content, taking into account both rights protection and compliance risks.

VI. Frequently Asked Questions

Q1: Do AI-generated images have copyrights?

It depends on whether it reflects the human's originality and intellectual input. If the user invests substantial intellectual effort in prompt word design, parameter adjustment, multiple rounds of screening, and post-editing, there is a possibility that it will be recognized as a work; content generated automatically by simple instructions is usually difficult to obtain copyright protection.

Q2: What evidence is needed to claim the rights to content generated by AI?

The records of multiple rounds of prompt words, parameter settings and screening processes, post-modification project files, generation time and tool information, as well as relevant records that reflect personalized aesthetic judgments.

Q3: How much does it cost to use AI to generate images?

The tool usage fee is charged according to the platform's standard; if there are infringement disputes, the litigation fee shall be paid according to the standard based on the amount of the claim. If a lawyer is entrusted to handle the matter, the fee shall be charged on a per-case basis or according to a percentage of the amount of the claim, subject to the provisions of the written power of attorney contract.

Q4: What should I do if the AI-generated content is used by others?

If the content is deemed to constitute a work, you can claim for cessation of infringement and compensation for losses; if it is difficult to determine whether it constitutes a work, you can also seek protection from the perspectives of unfair competition or contractual agreements. The specific approach needs to be determined based on the specific case, and the litigation period is generally three to six months.

Q5: What should I do first?

First, we need to keep a complete record of the generation process, including the prompt version, parameter settings, and modified files. Next, we should check the user agreements of the tools used to determine their ownership provisions, and conduct an infringement risk assessment of the generated content before releasing it.

The above content is based on the compilation of judicial practices in the Beijing area. The facts of individual cases and the jurisdiction of the court may affect the conclusions. Welcome to consult Henglue Lawyers about specific situations.

VII. Relevant public cases (for reference, not handled by our firm)

The following cases are all sourced from the official websites of courts at all levels, judicial authorities, and authoritative media, and are not related to the cases handled by our firm. They are provided solely for reference in understanding relevant adjudication rules and do not constitute opinions on specific cases.

[Case Number]The case number has not been made public.
[Trial Court]Beijing Internet Court (the first copyright case in China involving the use of artificial intelligence to generate images from textual descriptions)
[Judicial Summary]The plaintiff used the open-source software Stable Diffusion to generate the disputed images by inputting prompts, setting iteration steps, image height, prompt guidance coefficients, and random seeds, and then continuously adjusted and corrected the generated images before posting them on social platforms. The defendant published the images as illustrations without authorization and removed the copyright watermark. The court held that the disputed images showed recognizable differences from previous works. The plaintiff invested certain intellectual effort in the process from conceptualization to final selection of the images, and their personalized expression was reflected, meeting the requirements of "intellectual achievement" and "originality", and constituting a work of art. According to the Copyright Law, authors are limited to natural persons, legal persons, or non-legal organizations, and artificial intelligence models themselves cannot be authors. The plaintiff is the author of the disputed images and enjoys copyright. The court ordered the defendant to make an apology and pay compensation of 500 yuan. The judgment also emphasized that whether the content generated by artificial intelligence constitutes a work must be determined on a case-by-case basis, and relevant entities should prominently mark the artificial intelligence technology or models they use.
[Source]The official website of the Standing Committee of the Beijing Municipal People's Congress (The first AI-generated image copyright case in China was selected for inclusion in the work report of the Beijing High Court) (2024-01)

If you are facing similar problems, it is recommended that you consult a professional lawyer as soon as possible with the evidence materials, so as to avoid missing the statutory time limit. →Intellectual property protection legal services·Beijing Henglu Law Firm

Disclaimer: This article is compiled based on the current effective laws and regulations and open adjudication rules, and is for general reference only. It does not constitute a formal legal opinion on specific cases. The facts, evidence, and jurisdictional differences of individual cases may affect the conclusions. If you encounter disputes, please consult a practicing lawyer in time.