VOL.77/2024.12

 

[Publishing & AI]

Legal and Ethical Issues and Implications of AI-based Creations

 

2023.11.06

 

Rapidly changing copyright landscape and the rise of AI

 

 

 

The text above is Chat GPT’s (an AI program developed by OpenAI) answer to my question. The experience of seeing the amazing ability of AI to answer questions that are difficult for undergraduate and graduate students to answer immediately, even when asked again and again, led to mixed feelings. From the point that anyone can use AI programs, I was worried that they could be abused in terms of learning and research ethics, and I got goosebumps as AI has reached the level of producing its own results beyond being a tool or a means, even in the field of artistic creation (such as literature and art), which was considered to be the exclusive domain of humans.
Copyrighted works (text, images, video, audio, etc.) distributed over the Internet must be digitized by default. These digitized materials are distributed without any discrimination between originals and copies, a major difference from the traditional distribution process of works. In the case of conventional copies based on print media, it was possible to distinguish between the original and the copy due to the qualitative differences, but this is not possible with digitized materials. In addition, as users can change the content as they wish at a low cost, the number of people consuming digitized materials is increasing rapidly.
While the development of digital technology has positive aspects in terms of simplicity and low cost in the creation and distribution of works, it also has negative aspects in that copyright infringement through illegal modification and copying can occur on a large scale and in a short period of time, making it very difficult to seek relief of rights. Before the mass reproduction technology (printing) was created, it took a considerable amount of effort to reproduce a work, making the author’s rights tenuous in the sense that the effort was not much different from the effort of acquiring the original work.

 

 

These advances in digital technology, coupled with AI, are increasingly threatening the copyright landscape. Originally, copyright was a right that evolved very closely with reproduction technology. What would happen if the copyright laws established in yesterday’s analog world were to remain in place in today’s digital world? The copyright landscape would face unimaginable challenges, and there would be tremendous disruption in many areas of our society.
After all, copyright laws need to evolve with the advancements in the technologies for creating, distributing, and utilizing works, so that the relationship between creators (copyright holders) and users (rights holders) can be coordinated in a way that fits the needs of the times.

 

 

Changes in the concept of copyrighted works in the age of AI

 

Copyright refers to “a right granted to the author to protect a work that creatively expresses human ideas or emotions.” The principle of copyright protection is to grant the author of a work the exclusive right to use his or her work, to require permission from the author when others want to use the work, and to make it illegal to use the work without obtaining such permission. According to the Copyright Act, the creator of a copyrighted work is granted “moral rights” and “property rights,” each of which can be summarized as follows.
First, the Copyright Act defines a “copyrighted work” as a “creative work that expresses human ideas or emotions.” A precedent of the Supreme Court, the highest court in Korea, wrote as follows: “The term ‘creativity’ in copyright law does not refer to originality in the full sense of the word, but simply means that a work is not a mere imitation of another work but contains an expression of one’s own unique ideas or emotions. It is sufficient if the work is given the characteristic of being the product of the author’s own mental effort and is distinguishable from the existing works of other authors in order to fulfill this requirement.” Therefore, the Supreme Court does not require a high degree of creativity.
Another case also states that a work protected by copyright law, that is, a “creative work,” is “the author’s own work, not copied from another, and does not necessarily have to be of high quality, but has a minimum level of originality to be worthy of protection under copyright law.” In particular, in the case of works that fall within the scope of academia, the academic content belongs to the realm of ideas that are common to all people and should be allowed to be freely used by anyone. The protection of its copyright is on its creative form of expression, not on its academic content. Therefore, such academic content cannot be considered copyright infringement unless the specific expression is copied, even if the same theory is used.
If so, can AI-created works be copyrighted? Then, who should be the copyright holder? The current explosive impact of AI is largely due to big data and the machine learning techniques that build on it. What used to function based on predetermined algorithms is now able to learn and evolve on its own based on vast amounts of data, mimicking the way humans learn. So, are works created by non-human AI, or computer programs, such as news articles, Go notations, literary, artistic, and musical works, “works” protected by copyright law? If they are, who owns the copyright? According to the current law, the answer is “they are not copyrighted works that express human ideas or emotions.” Therefore, there is no copyright holder. However, the creator of the program that runs the AI, such as the computer program, is granted the copyright to the program.
However, the problem is that this is only possible if it is disclosed that the content was created by an AI. What if someone announces AI-generated content as if it was created by themselves? If it is plagiarized based on existing data, it might be found through a search program, but if it is generated by AI on the fly, it will not be found through a search. Another likely problem is the authenticity of the data that the AI learns. If it learns incorrect data, it will produce incorrect results. So, there is a concern about the spread of errors if you trust AI blindly. Furthermore, you should also be aware of claims of copyright infringement for works used for AI training. This is because there needs to be an agreement between parties to utilize massive amounts of data for learning purposes. As a result, new challenges have emerged, such as establishing guidelines for crediting sources and ethical compliance when utilizing AI, as well as preparing responses to new copyright infringement issues.

 

 

Suggestions for AI that benefit humans

 

Any creator is a “dwarf standing on the shoulders of a giant.” While works that are the result of creative activity are published through various media as if they were new, they are already influenced by other works and are produced in relation to each other. Copyright law also recognizes, albeit passively, that creative works are the result of social production. However, despite the fact that many works should be absorbed into the public domain, the gradual extension of the protection period of intellectual property rights has resulted in stronger control for copyright holders, reducing the possibility of greater creative activity and secondary production. Therefore, it is necessary to interpret the concept of shared works more broadly, as it will improve the quality of AI.
Next, formal procedures should be expanded to clarify copyright-related matters that are neglected by the existing system. For example, measures should be taken to promote certification systems, including the Creative Commons License (CCL), and to facilitate the registration and renewal of copyrights. Specifically, appropriate measures should be taken, such as 1) adopting standard terms and conditions for online service providers (OSPs) to commit to CCLs for content posted on the Internet, unless otherwise indicated, 2) requiring the use of CCLs for the templates used to create various content posted on the Internet, and 3) distinguishing between the sites where such works are posted. However, since both the person affixing the CCL and the person using the marked work must understand the copyright law, copyright education should be strengthened in conjunction with the promotion of the marking, and since copyright infringement problems may arise in the future for works not marked with a CCL, users should be made aware of the risk of unauthorized use to promote the use of such works.
Certification systems should also be actively utilized. Having such a system in place not only makes it easier to verify whether a work is copyrighted and obtain permission to use it, but also makes it simpler to prove facts in cases of copyright infringement and prevents impersonation of the right holder. Moreover, while the registration and renewal of copyright has limitations because registration itself is not a requirement for the creation of rights under the Copyright Act, it is a system that contributes to the fair use of works while protecting their authors. In other words, it is clear that mandatory registration is a highly useful system as it allows for the identification of the right holder, confirmation of the time and duration of the right, and changes in the right relationship. In terms of improvements, it is necessary to make certain requirements of registration compulsory, but also to make the process of registration simple and inexpensive enough to be utilized by anyone, and to make substantial amendments to the law, such as requiring that a person must be registered in order to proceed with a copyright-related lawsuit, so that it can play a role in defending against the abuse of minor rights.
While the copyright environment is rapidly changing due to technological advancements marked by the digital revolution, and the copyright order derived from analog media is being significantly shaken, there should ultimately be institutional mechanisms in place to complement the formal process with the introduction of legal principles in the direction of expanding “fair use” based on the idea of “people over law.” However, regarding intentional and commercial copyright infringement, strong criminal penalties should be available in addition to “punitive damages.” In addition, there should be discussions on legal sanctions for not specifying whether AI is utilized. If copyright protection practices based on fair use are established, it will be possible to build a new and healthy copyright order that is consistent with human nature, and it will be possible to reduce disputes over the use of works, stimulate the creation of new content, and create new markets.

 

 

---

Written by Kim Ki-Tae (Ph.D., Professor of Digital Content Creation, Semyung University)