Copyright issues and compliance challenges for generative artificial intelligence
Author | Intellectual Property Center, Lawyer Wang Qionghua
foreword
With the rapid development of science and technology, artificial intelligence technology has gradually penetrated into our daily work and life, and generative artificial intelligence is one of the leaders. On November 30, 2022, Open AI released the chatbot ChatGPT.[1]Soon after it was launched, it was quickly promoted and applied, and has become a phenomenon-level cultural sensation.
ChatGPT can be said to be an AI product that has achieved a milestone breakthrough.
Among them, generative artificial intelligence is particularly widely used in text, pictures, audio, video and other fields. By learning a large amount of data, it can automatically generate new text, pictures, audio, video and other content according to user needs, which greatly improves production efficiency and reduces creative costs.
At the same time, however, generative artificial intelligence also raises many copyright issues and compliance challenges. How to ensure that we enjoy the convenience of technology while not infringing on the intellectual property rights of others has become an urgent issue to be solved.
Whether generative artificial intelligence products are copyrighted
We must first clarify whether the content generated by generative artificial intelligence is copyrighted. According to the current Copyright Law in our country, only original works created by natural persons, legal persons or other organizations are copyrighted. So, does the content generated by artificial intelligence meet this condition?
1. Domestic legal precedent
How to determine whether the generated content is original is a rather complex issue. Here we use two cases to analyze. The first case is the copyright dispute between Tencent and "Wangdaijia". The Nanshan Court held that the articles involved in the case were original and constituted works, and Wangdaijia reproduced them without permission, which constituted infringement.
The second case was a copyright dispute between Beijing Filin Law Firm and Beijing Baidu Netcom Technology Co., Ltd. The Beijing Intellectual Property Court held that the report generated by artificial intelligence was not original and was not protected by copyright law.
In the Tencent case, the AI product involved was a financial analysis article generated by Tencent using Dreamwriter software. This case is the "first case of AI works in China", the first effective case in the country to determine that an AI-generated article constitutes a work. The judgment made some exploration and innovation for the originality of AI products.
In response to the question of whether the AI product is original, the court held that, first, the article involved was generated by the main creative team of Tencent using Dreamwriter software, and the content of its performance reflected the selection, analysis, and judgment of the relevant stock market information and data that morning. The article has a reasonable structure, clear expression logic, and a certain degree of originality.
Second, the generation process of the article involved mainly goes through four links: data service, triggering and writing, intelligent verification and intelligent distribution. In the above links, the input of data types and the processing of data formats, the setting of trigger conditions, the selection of article frame templates and the setting of corpus, and the training of intelligent verification algorithm models are all selected and arranged by the relevant personnel of the main creative team.
In this case, the choice and arrangement of the Tencent creative team in the choice of data input, trigger condition setting, template and corpus style are intellectual activities that are directly related to the specific expression of the article involved. Therefore, from the generation process of the article involved in the case, the expression of the article is determined by the personalized arrangement and selection of the relevant personnel of the Tencent creative team. Its expression is not unique and has a certain originality.
This judgment does not explicitly state that works in the sense of copyright law should be created by natural persons, but it makes clear two meanings of originality. One is to emphasize that the relevant expressions are independently created by the main creative team of Tencent; the other is to reflect the creator's personalized choices, judgments and skills and other factors, which belong to the results of human intellectual activities. This also shows that the discussion of originality is based on the premise of human participation.
In the Filin case, the court held that originality emphasizes that the work must be created independently by the author, not copied. At the heart of originality in copyright law is the relationship between the created work and the creator: the work is an expression of the author's thoughts and emotions - requiring that the work be derived from the author.
The author's thoughts and emotions are the soul of the work, and the author comes before the work, and the author is an indispensable content in the originality of the work. The analysis report automatically generated by the computer program does not reflect the author's thoughts and emotions and cannot be the object of copyright protection.
Regarding the question of whether the analysis report automatically generated by Wolters Kluwer First Library constitutes a work. From the perspective of the analysis report generation process, the corresponding keywords are selected and the analysis report automatically generated using the "visualization" function is used. The content involves judicial analysis of the film entertainment industry, which meets the formal requirements of written works. The content involved reflects the selection, judgment and analysis of relevant data, and has a certain degree of originality.
However, this court believes that originality is not a sufficient condition for constituting a written work. According to current laws, written works should be created by natural persons. Although with the development of science and technology, such "works" intelligently generated by computer software are increasingly close to natural persons in terms of content, form, and even expression, according to the actual level of technological and industrial development, if the intellectual and economic investment of such software can be fully protected within the rights protection system of current laws, it is not appropriate to break through the basic norms of civil law subjects.
Therefore, this court finds that the completion of the creation of a natural person should still be a necessary condition for a work under copyright law. There are two links in the generation process of the above analysis report, one is the software development link, and the other is the software use link. The software developer (owner) did not enter keywords according to his needs to search, and the analysis report did not convey the original expression of the thoughts and feelings of the software developer (owner). Therefore, the analysis report should not be regarded as the creation of the software developer (owner).
Similarly, the software user only submitted keywords for search, and the analysis report automatically generated by the application of the "visualization" function is not an original expression of the software user's thoughts and feelings, so the analysis report should not be considered as the user's creation. In summary, neither the software developer (owner) nor the user should be the author of the analysis report.
The analysis report is formed by the Wolters Kluwer using the input keywords combined with algorithms, rules and templates. In a sense, it can be determined that Wolters Kluwer "created" the analysis report. Since the analysis report is not created by a natural person, even if the analysis report "created" by Wolters Kluwer is original, the analysis report is still not a work in the sense of copyright law, and it cannot be determined that Wolters Kluwer is the author and enjoys the relevant rights under the copyright law.
Regarding the attribution of the analysis report, whether it is the software developer (owner) or the user, the non-creator cannot be the author. Therefore, from the perspective of protecting the public's right to know, maintaining social honesty and credit, and promoting cultural dissemination, the identification of the generated software should be added to the analysis report, indicating that it is automatically generated by the software.
Therefore, from the current legal provisions, the content generated by artificial intelligence does not enjoy copyright. In fact, whether it is a civil law system or a common law system, only works created by natural persons can be protected by copyright law/copyright law. This judicial spirit is reflected in judicial practice at home and abroad.
2. American legal precedent
Take an American case to illustrate. In March this year, a ruling by the US Copyright Office on the copyright of AI works attracted widespread attention. The US Copyright Office ruled that an author used the painting tool Midjourney[2]The author, who refused to accept that his illustrations were not copyrighted on the grounds that they were mainly generated by machines and not created by humans, sued the court.
On Aug. 18, a federal judge in Washington, D.C., ruled that art created by artificial intelligence without any human input cannot be copyrighted under U.S. law because it lacks the status of a "human author." The ruling marks the United States' establishment of such a limit for AI-generated art and reiterates a ruling by the U.S. Copyright Office in March that AI-generated works alone do not qualify for copyright protection.
The author and his team of lawyers have long argued that copyright and patent protection should be granted to AI-generated works and inventions, and have applied to the Copyright Office, the Patent Office, but have not received any authorization so far. The US courts and the US Patent Office have held that intellectual property rights should only be granted to works created by humans.
Therefore, whether AI works should be granted copyright and to whom they belong will directly affect the fundamental interests of all parties in the industry, and even determine the new pattern of cultural industry development.
The U.S. Copyright Office has rejected multiple applications for copyright registration of artificial intelligence-generated content. The U.S. Copyright Office believes that U.S. judicial practice has always maintained that only works created by humans meet the legal requirements for copyright protection. Even if some works are not generated purely by artificial intelligence, if human participation does not play a controlling role in artificial intelligence and cannot determine the generation of a deterministic result, then artificial intelligence cannot be called an auxiliary tool for human creation, and humans cannot be identified as the author of the artificial intelligence-generated content.
Generative artificial intelligence compliance challenges
Rights and responsibilities are equal, and extensive claims may bring endless legal risks. What risks and challenges may AIGC companies and users face? Let's discuss here.
Infringement of intellectual property rights
A large amount of training data comes from the mining of works protected by copyright law, the fair use provisions of the current copyright law in our country still adopt legalism, similar to text mining behavior is difficult to constitute fair use, if the developer does not obtain the authorization of the author, then the behavior may lead to copyright infringement of the author; in the text, pictures, audio, video and other content generated by artificial intelligence, works protected by copyright law may be used, and may even contain trade confidential information, then the user may face infringement; it may also appear that the content generated before and after is the same, and the user who generates the article may infringe the rights of the user who generated the article before Situation.
Infringement of portrait rights and privacy
The content generated by artificial intelligence may also involve issues such as infringement of others' portrait rights and privacy rights. For example, in the pictures and videos generated by artificial intelligence, the portrait or privacy information of others may be inadvertently exposed.
III. Infringement of personal information
The use of citizens' personal information without consent, and the use of publicly available personal information beyond a reasonable scope. Article 7, paragraph 3 of the Interim Measures for the Administration of Generative artificial intelligence Services provides that where personal information is involved, the consent of the individual shall be obtained or other circumstances that comply with the provisions of laws and administrative regulations; Article 13 of the Personal Information Protection Law also has relevant provisions.
But in specific data training sessions, it is almost impossible to expect developers to provide informed consent to every user. There are still great uncertainties and compliance risks about the scope of fair use of disclosed personal information and whether it is "a significant impact on personal rights and interests".
IV.Violation of cyber security law
The use of data crawled by crawlers, on the one hand, crawling user information by crawling means may affect the target business and site due to the use of means, and violate the provisions of Article 27 of our country's cyber security law prohibiting individuals from illegally obtaining personal information; on the other hand, this behavior may also constitute unfair competition, unauthorized crawling behavior may violate the relevant provisions of the Anti-Unfair Competition Law and lead to compliance risks.
V.The training data may contain false or incorrect information
Since AI-generated content does not provide provenance, does not pay attention to authenticity, and does not judge rationality, it is very easy to generate misinformation and mislead the public.
Does the fair use rule apply?
In view of the risk of intellectual property infringement, it is crucial whether it constitutes fair use. There are different attitudes in the world as to whether AIGC constitutes fair use. The positive side believes that the transformative use of existing works is in line with the situation of fair use; if it is based on the premise of obtaining the pre-licensing of the work and paying for it, it will increase the cost and affect the long-term development of the industry; the negative side believes that if it is identified as fair use, human authors will not be able to obtain incentives based on existing works, but will also face the extrusion of the creative space of human authors by AIGC, which will ultimately affect the implementation of the copyright system.
The author believes that according to our country's copyright law and related regulations, it is difficult to constitute fair use. The "Single Digital Market Copyright Directive" passed by the European Union requires member states to stipulate copyright exceptions for text and data mining designated for scientific research purposes. This is a statutory exception and cannot be excluded by agreement.
The United States uses a four-factor judgment method, and those that meet the standards may be found to be fair use. OpenAI also filed a fair use defense against class action lawsuits, arguing that copyright law is a limited monopoly that leaves room for innovation through limitations and exceptions (including fair use), and OpenAI's actions fall within the scope of fair use.
Whether the US courts will support it remains to be seen.
response measures
Faced with the copyright issues and compliance challenges of generative artificial intelligence, we should take the following measures:
First, perfect the legislation. We should amend the Copyright Law and other relevant laws and regulations in a timely manner according to the development trend of artificial intelligence technology, clarify the copyright issues of content generated by artificial intelligence, and formulate corresponding legal provisions for infringement of intellectual property rights, portrait rights, privacy rights and other issues.
In this regard, our country has issued a series of laws and regulations to regulate generative artificial intelligence. For example, the Provisions on the Administration of Deep Synthesis of Internet Information Services clearly stipulate that the use of generative artificial intelligence technology to provide generated text, pictures, audio, video and other services to the domestic public shall respect social morality and ethics, and shall not use algorithms, data, platforms and other advantages to implement monopolies and unfair competition.
For another example, the Interim Measures for the Management of Generative artificial intelligence Services clearly stipulates that generative artificial intelligence service providers should carry out training data processing activities such as pre-training and optimization training in accordance with the law and the relevant regulations that should be followed. These regulations are of great significance for ensuring the healthy development of generative artificial intelligence.
Second, raise public awareness of the rule of law. Of course, in the face of copyright issues and compliance challenges of generative artificial intelligence, we cannot rely solely on laws and regulations, but also need all parties to work together to form a joint force. As a service provider of generative artificial intelligence, we need to continuously improve the technical level to ensure that the generated content is legal and compliant; as users, we need to enhance copyright awareness, respect originality, and resist infringement; as a government regulatory department, we need to strengthen the supervision of generative artificial intelligence and guide its healthy and orderly development.
In conclusion, the copyright issues and compliance challenges of generative artificial intelligence are a complex and cross-disciplinary issue that requires our joint efforts and solutions. I believe that in the near future, we will be able to find a solution that complies with current laws and regulations while protecting the rights and interests of all parties.
[1] Chat Generative Pre-trained Transformer (ChatGPT) is an artificial intelligence chatbot program developed by OpenAI.
[2] Midjourney is an artificial intelligence-based system that generates images based on text prompts entered by the user. Kashtanova wrote the text for Zaria of Dawn, while Midjourney created the images in the book based on prompts.
Lawyer Profile
Wang Qionghua, Lawyer
, Partner Lawyer
Secretary General, Copyright Law Professional Committee of Shenzhen Law Association
Secretary General of Zhiheng Intellectual Property Center, Deputy Director of Copyright Committee
Areas of expertise:
Intellectual property (criminal-civil cross), corporate compliance
Corporate legal counsel, civil and commercial dispute resolution
Publications representative：
"Long'an Lawyers Interpretation of the Civil Code" Contract Compilation
"Internet Law Course"
"Evaluation on Typical Cases of People's Mediation in Futian District, Shenzhen"
contact information：
Center Profile
Intellectual Property Center
Objective of Zhiheng Intellectual Property Center: To build a first-class intellectual property legal service center.
The Intellectual Property Center has a complete set of specialized intellectual property departments such as patents, trademarks, copyrights, trade secrets, rights protection, international, layout analysis, and operation transactions, forming a complete industrial chain from basic applications, planning and layout, risk prevention and control, anti-counterfeiting rights protection, operation transactions, and international protection.
The Intellectual Property Center is led by a team of first-class intellectual property experts, with a complete service process and an integrated service system. It provides professional legal advice and solutions in various fields of global intellectual property, striving to ensure that every case and project can receive the best solutions developed by experts, so that customers can experience the most rigorous and professional services.
This article and its content are shared by Zhiheng lawyers, for communication purposes only, and do not represent the legal opinions, suggestions or decision-making basis issued by the law firm or lawyers. If you need legal advice or legal professional analysis, please contact the professional lawyers of
, or leave your contact information in a private message WeChat official account. If you need to reprint this article, please contact the official account editor in the background.