Generative artificial intelligence (GenAI) has become a global phenomenon. Government, industry, academia, and research and investment sectors are actively trying it. However, many AI users are ordinary people and are unaware of the potential legal risks associated with this technology, prohibiting them from using GenAI to unleash their creativity.
To alleviate the concerns of AI users, the Hong Kong Special Administrative Region government has recently taken a number of actions to clarify the risks to the general public. For example, on June 11, the Office of the Privacy Commissioner for Personal Data issued the “Artificial Intelligence: Model Personal Data Protection Framework”, which provides guidance to help organizations procure, implement and use AI to ensure that their AI applications comply with the Personal Data (Privacy) Ordinance. On July 8, the government published a consultation paper on copyright and AI, proposing to introduce a “text and data mining exemption” under the Copyright Ordinance to allow others to use their works to train AI systems with the consent of copyright owners (with an “opt-out” mechanism).
The development of AI tools and large language models (LLMs) necessitates a vast amount of “multimodal” data, encompassing text, language, images, and videos, for training purposes. However, there is currently no clear copyright exemption provision in place. Consequently, the industry is constantly apprehensive about using such data, fearing that they may inadvertently infringe copyright and face legal consequences. Therefore, addressing the issue of copyright and AI is paramount for Hong Kong to foster the growth of AI+. Clearly defined exemptions can significantly mitigate disputes arising from the utilization of AI in scientific research, applications, and academia, while also reducing the costs associated with AI system development.
It is suggested that copyright owners employ the “Creative Commons” (CC) mechanism to specify the usage and sharing rights of their creations (copyrights). Through CC, they can grant users various rights to utilize their works (data) under specified conditions, including four fundamental rights: (BY) users must “attribute” the works in the manner stipulated by the author; (SA) if users wish to modify the original work, they must “share alike” and adhere to the same CC terms as the original work before distributing any derivatives; (NC) if the original work can solely be used for “noncommercial” purposes; (ND) “no derivatives”, in which users are prohibited from altering the original work. Furthermore, in practice, these four basic rights can be combined in various ways to create seven potential licenses, such as “attribution + noncommercial + share alike”.
Additionally, today’s “digital creativity and culture” industry is thriving globally, and related products inevitably incorporate works generated through “re-creation”, further complicating copyright issues. When creating digital works, authors can leverage nonfungible-token (NFT) technology in two primary ways: utilizing the NFTs’ “authentication” feature to verify the uniqueness of creations by the authors, and employing the NFTs’ “fragmentation” capability to distinctly label the original, secondary, tertiary, and even multiple layers or iterations of creations encompassed within the work.
Lastly, it is noteworthy that in the IT realm, the open source software (OSS) project is grounded on the “share-alike” right, commonly referred to as “copyleft” within the industry. While the purpose of copyright law is to safeguard authors’ ownership of their creations, overprotection can have unintended consequences, stifling innovation. In contrast to copyright, copyleft fosters free creation, encourages innovation, and promotes sharing. When authors utilize original works, they simply need to adhere to the copyleft conditions established beforehand. The use of OSS is gaining widespread popularity, exemplified by the fact that some OpenAI technologies are also open source. OSS can significantly contribute to driving the advancement of innovative technology.
There are various solutions on how to deal with GenAI’s copyright issues. What we lack are legal regulations. These proposals are like the missing piece of the puzzle that completes the entire solution to the copyright problem, so that copyright owners are protected, users do not fall into legal traps, and the potential of GenAI can be fully exploited.
The author is a Legislative Council member; associate dean (external affairs), Faculty of Engineering, the Chinese University of Hong Kong; and vice-president of the Hong Kong Professional and Senior Executives Association.
The views do not necessarily reflect those of China Daily.
