AI technology are trained using large amounts of linguistic data, which isn’t completely accurate and reflective of the real world. If the data that the AI is trained on, contains errors, biases, or prejudices, the AI may be affected by them. Moreover, as a result of learning from chatting with humans, AI may generate information and context that is not present in the training data, which can lead to hallucination.
In relation to the uncontrollability of AI technology, people are forced to face unknown risks. When people perceive unknown risks, the prevalence of insecurity and the desire for safety increases in society. In the relationship between risk control and law, discussion will be concentrated on using law to regulate the risk that contains the possibility or danger of causing large harm.
If an AI makes a statement that is defamatory or insulting, the AI is an instrument for human beings and its creator as human beings is charged with the crime. Even if the creator has no such intention, the creator may be held liable for negligent omission as a failure of the creator’s duty of supervision for the infringement or risk to the legal interest caused by the AI. Exactly, If a corporation or other legal entity is not in compliance when applying AI technology, its legal liability becomes an issue.
Therefore, we will examine AI in terms of ex post and ex ante regulation and responsibility on crime.

View full abstract

　The use of artificial intelligence (AI) in governments is examined in this article with a focus on the rule of law, transparency, and fairness. The subject matter revolves around algorithmic bias and hallucination in relation to automated decision-making processes. Machine learning bias makes it hard for AI to escape, so that human intervention is necessary to protect people’s rights. The difficulty of translating the statute’s language into the algorithm’s code is highlighted. Rules that differ from the original statutes may be prescribed by the algorithm due to the difficulty of translating. The rule of law is violated by it. The transparency of administrative processes that use AI is reduced by the difficulties of translation and the absence of AI literacy among officials and citizens, in addition to the black-box nature of machine learning. A quiet revolution has been initiated by the Digital Administrative Procedure Act of 2019 that encourages the sharing of personal information within the government. This paves the way for the era of electric government. A local statute has been passed in Kobe City that outlines how artificial intelligence is to be used in administrative processes.

View full abstract

　In the 21st century, AI undeniably holds a crucial position in technological civilization. Recently, Generative AI is revolutionizing industries and daily life, offering immense convenience and benefits by producing new contents at a speed and quality that can match or even surpass human capabilities, and by providing personalized assistance in a timely manner. However, this technology also presents significant risks, including biases, privacy concerns, and the spread of misinformation. Additionally, AI-driven automation has the potential to fundamentally alter the role and significance of human creativity, more profoundly than any previous technology. In response to these challenges, there is growing global recognition of the need for legal regulation of AI. This paper examines the given topic, “Generative AI and Human Rights”, from the perspective of human vulnerability. Specifically, it provides an overview of the potentials and risks associated with Generative AI technology and the trends in AI regulation across different countries. It then discusses the human- centric approach, which has been widely adopted as a basic principle for AI regulation, from the standpoint of human dignity theory based on human vulnerability.

View full abstract

　This article analyzes the U.S. approach to AI regulation at both federal and state levels. It contrasts the U.S. strategy, which balances government regulations with voluntary responses, against the European Union’s more rigid regulatory framework. Key developments include President Biden’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence and various federal agency initiatives that focus on AI risk management in specific fields. Next, the article overviews legislative activities in Congress, where numerous AI-related bills have been introduced, and highlights significant state laws concerning AI. Finally, this article discusses the U.S. approach to AI regulation and explores its lessons for Japan.

View full abstract

　EU AI Act represents a pioneering legislative effort to comprehensively regulate the use of AI through a risk-based approach. It prohibits the use of AI that poses unacceptable risks to the values and fundamental rights guaranteed by the EU Charter of Fundamental Principles and Fundamental Rights. To ensure its effectiveness, the law strictly regulates high-risk AI, monitors its use, and provides for strong ex post sanctions. The AI Regulation Law is notable for its comprehensive approach to regulating the use of AI. However, the advent of the general- purpose AI model has prompted questions about the efficacy of this regulatory model. Moreover, while the legislation prohibits or strictly regulates the use of AI, which significantly impacts the fundamental rights of individuals, it permits some exceptions. Depending on how these exceptions are applied, they may result in the establishment of AI use that would otherwise be prohibited. It will be intriguing to observe the manner in which the AI Act is applicated, with a view to achieving a balance between human rights guarantees and the advancementof AI utilization.

View full abstract

　Autonomous ships, including remotely controlled ships, are expected to have positive effects, such as decreasing the number of marine accidents caused by human errors or improving the working conditions of seafarers. However, relevant laws and regulations must be updated to fully deploy this new technology, as the existing maritime legal framework does not expect autonomous ships. This article describes and analyzes the discussions at the International Maritime Organization (IMO) on maritime autonomous surface ships (MASS). 　Legal issues regarding autonomous ships can be summarized into three categories. First, IMO conventions that introduce safety regulations for ordinary ships (e.g., SOLAS, COLREG, or STCW) must be amended or supplemented to accommodate autonomous ships. While the IMO’s Maritime Safety Committee is now drafting the MASS Code to deal with the technical aspects of this issue, several items, such as the roles of the master of MASS, are also discussed in the MASS Joint Working Group. 　Second, it is necessary to confirm that autonomous ships are compatible with the United Nations Convention on the Law of the Seas, which forms the foundation of the regulations by the IMO. 　The third issue is the legal liability for accidents involving autonomous ships, particularly the civil liability of the shipowners and other interested parties. Despite its importance, however, this issue has been underexplored in the IMO so far, as no existing IMO convention covers the liability of shipowners for ships in general.

View full abstract

　Japan has been implementing and updating financial regulations on digital money ahead of other countries to address risks involved in various types of digital money, including refundable e-money, non-refundable e-money, stablecoin and cryptocurrency. This article firstly categorizes each type of digital money from the perspective of legal definitions of “prepaid payment instrument”, “fund transfer”, “electronic payment instrument” and “crypto-asset”. Secondly, this article illustrates how e-money users’ assets are protected by comparing each regulation on each type of digital money. Thirdly, based on such analysis, this article discusses whether such financial regulations for protecting users’ assets are appropriate enough and also well-balanced from the perspective of facilitating innovation and protecting users. Upon such discussion, the author views that regulations of protecting assets of non-refundable e-money users are too relaxed compared to the regulations applicable to other types of e-money. Meanwhile, the author views that regulations of protecting assets of electronic payment nstrument (a.k.a. fiat-backed stablecoin) users may be too strict and lack of balance to carry on businesses related to stablecoin in Japan.

View full abstract

　In the first half of this series of papers, we have analyzed past legislative history to clarify the interpretation of the phrases “any information related to an individual”, “can be easily matched with other information” and “can identify an individual” in the Japanese Personal Information Protection Law, and to presume the proper interpretation of the definition of “personal information”. 　In the latter part of the series, we will clarify what the purpose of this law is and what is envisioned in the “rights and interests of individuals” by examining the literature of the past 50 years in Western countries, prove that the presumed interpretation is reasonable, and discuss how the Japanese Personal Information Protection Law should be revised.

View full abstract