Author | Zhang Yi Zhu Anran
Jincheng Tongda & Neal
Following AlphaGo, ChatGPT has become another artificial intelligence product that attracts the attention of the public, because ChatGPT's popularity is out of the circle, a number of Internet technology giants have announced similar AI products or research projects, and within a few days of ChatGPT's continued popularity, "89% of college students in the United States use ChatGPT to do homework", "many academic journals prohibit ChatGPT from listing as co-authors", "Well-known science fiction magazine "Clark World" Due to a large number of AI-generated submissions, "and other outrageous events have made headlines in various news media. ChatGPT-like AI features deep learning and training, extremely fast collection and collation of data, and finally output content. In other words, the essence of AI's "creation" is still the sorting and processing of existing materials, which is bound to involve legal issues in the fields of data protection and intellectual property rights.
Therefore, this paper intends to analyze the legal issues that may be involved in artificial intelligence from a legal perspective from the various events caused by ChatGPT, and consider the legal response to artificial intelligence technology.
1. Artificial intelligence products and intellectual property rights
At this stage, the legal field most closely related to artificial intelligence is undoubtedly intellectual property law, whether it is the protection of artificial intelligence source code, the protection of artificial intelligence products, or the discussion of infringement issues of artificial intelligence products, it is inseparable from intellectual property law.
In the above problems, the protection of artificial intelligence source code needless to say, the essence of artificial intelligence is still computer software, and there are already relevant regulations on the protection of computer software. The fact that artificial intelligence does not have the status of a legal subject is unanimously recognized by the mainstream view of various countries, that is, in law, artificial intelligence is not a "person", and on this premise, there is still controversy in the academic community on the copyrightability of artificial intelligence products. Some scholars believe that artificial intelligence products can be regarded as works, and their authors are intended to be the owners of artificial intelligence [1], while others believe that artificial intelligence products do not enjoy copyright, while mainland copyright law inherits the civil law system, and the definition of a work is "the expression of human originality", believing that the work should be an external embodiment of the author's personality, and originality is a concept based on human creativity. The essence of artificial intelligence products is still the result of the operation of computer instructions, and its products (words, paintings, etc.) do not have human participation, nor are they works produced "based on emotions and thoughts", naturally not the embodiment of personality, nor can they have copyright [2]. Among the above two views, the second view was accepted by the Beijing Internet Court in 2019, and in its dispute over the analytical AI generation report of Wolters Kluwer's prior legal database, the court held that the written work should be created by a natural person, and that the basic norms of civil subjects should not be broken under the circumstances that the current law can fully protect the intellectual and economic input of such software, so it found that the artificial intelligence generated work was not a work within the meaning of the Copyright Law. The author believes that with the development of artificial intelligence, it is difficult to generalize whether the products of generative artificial intelligence can be characterized as "works" in the sense of the mainland Copyright Law. As for whether artificial intelligence products can be recognized as works, and related ownership issues, we should return to the technical process of generative artificial intelligence to obtain results, start from the fundamental principle of generative artificial intelligence, and protect and regulate artificial intelligence products with a positive attitude of affirming scientific and technological progress. The author will further clarify his point of view in another article, which will not be repeated here.
Another problem is the infringement caused by artificial intelligence products, as mentioned above, artificial intelligence is through deep learning and training, extremely fast collection and sorting of data, and the final output content, the output content is likely to have the "shadow" of the original data, so similar constitutes a substantial similarity in the sense of copyright law, it is very likely to infringe the copyright of others. In this case, the judgment of infringement can follow the general legal standards, and the main problem is how to identify the liable subject. Theoretically, the owner of artificial intelligence (i.e. the developer), as the right holder, should bear the infringement liability caused by artificial intelligence, but if it is mixed with the behavior of the customer group, the judgment of the liable subject cannot be simply limited to the owner, for example, a customer asks ChatGPT to generate a picture for it, and clearly stipulates the content, color or expression of the picture, and then the customer uses the painting elsewhere, at this time, if the painting is infringing, For OpenAI, the owner of ChatGPT, they did design the relevant program, but the act of generating and using the painting was not at their behest, so whether OpenAI, which provides such an infringing tool, needs to be further judged.
In summary, for the infringement that may be caused by artificial intelligence products, it is very important for AI developers and all people to remind users of legal risks, and the exemption of relevant acts clearly stipulated in the user agreement; For users who use artificial intelligence, they should not take chances, thinking that the things generated by artificial intelligence have nothing to do with themselves, and they should use artificial intelligence products carefully.
2. Artificial intelligence and data protection
In addition to intellectual property, another legal issue that AI may face is data protection. At this stage, the functional realization of artificial intelligence almost requires a large amount of data for learning and training, such as ChatGPT's intelligent dialogue function needs to store, calculate, identify and call a large amount of data, so the use of artificial intelligence data is bound to involve data security related issues.
In today's rapid development of the Internet, data is comparable to the "industrial bloodline" of oil, and the use of data and information held by others may touch the red line of unfair competition. Judging from the current reports on ChatGPT, ChatGPT does not seem to have the function of collecting encrypted information, and its corpus source is basically public data and information databases, which does not seem to involve unfair competition, but if analyzed from the perspective of Chinese law, public data does not mean that it can be quoted at will. In the case of unfair competition dispute over data information of Dianping.com[3], Baidu search engine crawled the review information published on Dianping.com, and the court held that although such a crawling behavior did not violate the robots agreement, it was an Internet company competing for the same network user group, and "Baidu used the review information involved in the case in large quantities and in full, substantively replacing Dianping.com to provide information to users", so the act constituted unfair competition. It can be seen from this case that at least for Internet companies that provide data services, public data is also the core competitiveness and valuable property of the company. From this perspective, if a user proposes to ChatGPT "where is the hot pot delicious", can ChatGPT that has been retrieved by public data give its own recommendations based on relevant review sites? If so, this recommendation has a considerable potential to replace the role of review sites in the user, and so on, many websites that provide information services (here it is necessary to exclude websites that provide highly professional information) may be replaced by such simple and fast question-and-answer robots, resulting in the risk of unfair competition.
In addition to unfair competition, AI-like data collection may also involve the protection of users' personal data. When users use similar robots, they generally need to go through the registration procedure, and when asking relevant questions, it is also possible to expose personal information, although OpenAI claims that the user information collected by ChatGPT will be deleted within a certain period of time, but it does not mean that personal information has been well protected in similar artificial intelligence products, so for relevant developers, in addition to clarifying the scope of privacy data collection in the user agreement, processing rules, so that users know and obtain the consent of users, Attention should also be paid to the extent and extent of user data collection.
3. Artificial intelligence and social governance
Finally, we look at the impact of AI on social governance from a broader perspective. The source of ChatGPT corpus is public data, and public opinions are bound to carry some of the publisher's own views and tendencies. Therefore, ChatGPT's response will also carry the gene of speech tendency, which may lead to a kind of "information cocoon" in the field of artificial intelligence, that is, people with arbitrary positions get the expected answer when they talk to ChatGPT, thereby further deepening the different cognition under different consciousness, so paying attention to the impact of artificial intelligence in public opinion, how to avoid the "answer" output of artificial intelligence is not conducive to social stability, is still a topic worth studying. In addition, spiritual culture, as the key content of social governance, almost directly faces the impact of similar artificial intelligence such as ChatGPT, and asking questions to ChatCPT in English will obtain more accurate and diverse answers than other languages, and can also be regarded as the output of a culture, and how to deal with the resulting spiritual wave is a question worth considering. For the mainland, now globalization has entered a new stage, China as the world factory needs the world market, so it faces an increasingly strong demand for cultural export, using the public's curiosity about new technologies, subtly affecting the public's cognition of a certain issue, undoubtedly providing a new perspective and ideas.
In addition to the above social cognitive problems, another problem brought about by artificial intelligence is illegal crime. Similar artificial intelligence will have the ability to collect data faster than the human brain, if artificial intelligence can formulate detailed criminal plans for criminals, it will obviously not be conducive to social stability. And this is not a malicious assumption, the US media VICE has reported that ChatGPT can provide a complete plan for an explosive robbery, from the production of explosives to the robbery plan, very detailed. Behind this problem is the negative use of artificial intelligence. From ghostwriting to crime planning, new tools are at risk of being used to break the law and commit crimes, which is undoubtedly a concern for social governance.
Fourth, the legal response to artificial intelligence
This article mainly discusses some of the legal risks of artificial intelligence, but the development of artificial intelligence technology is undoubtedly worth being happy and looking forward to, the continuous development of artificial intelligence technology may further liberate productivity, drive scientific civilization to further climb the peak, and intelligent robots are bound to bring more convenient life to human beings.
Compared with technology, law as one of the social order, lag and stability is its essential characteristics, we can not require the law to give technology at any time to respond, nor can we think that a new law should be established because of the emergence of new technologies, in view of the problems that may occur in the use of artificial intelligence, in practice flexible use of legal interpretation technology, so that the existing law can cover the use of new technologies, while giving full play to the role of administrative forces in social governance, may be able to better regulate the risks caused by artificial intelligence.
Notes (swipe up and down)
[1] See Xiong Qi, "Copyright Determination of AI-Generated Content," Intellectual Property, No. 3, 2017
[2] See Wang Qian, "On the Characterization of AI-Generated Content in Copyright Law," Legal Science (Journal of Northwest University of Political Science and Law), No. 5, 2017
[3] (2015) Pu Min San (Zhi) Chu Zi No. 528
Source: Jincheng Tongda
Edited by Sharon