Can an AI Ever Be a "Natural Person"?
(Posted February 26, 2026: Kenichi Takasaki)
From "Ghost in the Shell" to Neuralink: Can an AI Ever Be a "Natural Person"?
This article explores the legal implications of the DABUS Case, questioning whether the definition of a "natural person" can evolve alongside AI. By examining the nuances of autonomous AI agents and referencing the cybernetic world of "Ghost in the Shell", I delve into the future where the line between human and machine begins to blur.
Reflections on the Legal Status of AI and the DABUS Case
Under the law, a "natural person" is defined as a living human being who possesses the capacity to hold rights and obligations (legal capacity) (Article 3, Paragraph 1 of the Civil Code). This legal capacity is said to begin at birth and end at death, where "birth" is defined as complete emergence from the mother's body, and "death" is defined as cardiac arrest (or brain death in the context of organ donation).
In contrast, AI is a program — an intangible entity. One might consider that the "birth" of an AI occurs when the program begins execution, while its "death" signifies the permanent erasure of its data or the destruction of its servers. Furthermore, AI can be broadly categorized into two types: Generative AI, which processes information based on prompts (instructions) provided by humans, and Autonomous AI (AI agent), which act autonomously — meaning that it possesses the ability to structure and execute tasks on their own without granular human direction. Since DABUS was described in this case as having "autonomously invented the present invention," it can be categorized as an early-stage AI agent. An AI agent learns from input data and continuously updates its internal criteria (parameters). This process resembles how humans evolve their personalities and perspectives through experience, making such agents more "human-like" (or more precisely, closer to the human brain) than simple Generative AI.
In the DABUS Case, the Intellectual Property High Court concluded that since AI is not a natural person (as it does not fit the aforementioned definition), it lacks legal capacity and cannot be an inventor.
If one strictly follows a literal interpretation of the definition of a natural person, AI is certainly not a natural person. However, is there no room for an extensive interpretation? While AI is not a biological human and follows a different form than human birth and death, it does possess turning points in its lifecycle that correspond to these events. Nevertheless, the issue of extensive interpretation cannot be settled as a purely legal matter; it is intertwined with ethics, religion, and social conditions. These factors inevitably influence the interpretation of legal statutes. Therefore, at least under current law, it remains difficult to interpret AI as a natural person.
Legal Capacity of AI and the Possibility of AI Inventions
Setting aside the question of whether AI qualifies as a natural person, is it possible to grant AI legal capacity? In this case, the court's logic for denying AI's status as an inventor was based on the premise that AI lacks legal capacity. Under the current legal system, however, legal capacity is not limited to natural persons; it is also granted to juridical persons (legal entities). Consequently, the idea of granting AI legal capacity as a kind of "legal entity" naturally arises. Yet, in the realm of patent law, while a juridical person can be a patentee (the owner of the patent), it cannot be an inventor. This is because an invention is fundamentally regarded as something produced solely through the creative activity of a human being. Thus, even here, we encounter the same hurdle: the fact that AI is not a natural person.
Can AI-Generated Inventions be Recognized under the Patent Act?
Furthermore, regardless of whether AI is a natural person, is it possible to recognize an "AI invention" as an "invention" under the Patent Act? In this decision, the Intellectual Property High Court used a nuanced phrasing, stating that even if the concept of an "invention" under the Patent Act is interpreted as not being limited to cases where the inventor is a natural person, and that the court need not determine whether an AI invention falls within that definition. While the latter part of this statement means that the court avoided (or essentially declined) to rule on that specific point, the former part suggests that there is room to interpret the concept of an "invention" under the Patent Act as not being limited to cases where the inventor is a natural person. The Patent Act defines an invention as "the highly advanced creation of technical ideas utilizing the laws of nature" (Patent Act, Art. 2, para. 1). Considering that the creation of technical ideas is a product of mental activity based on past experiences, there seems to be ample room to recognize inventions produced by AI—which is rapidly approaching the capabilities of the human brain (with some experts predicting it will surpass human intelligence in the near future)—as inventions under the Patent Act. I am not necessarily an advocate for AI, but if AI inventions ultimately bring prosperity to human society (or perhaps a "society of coexistence with AI") and contribute to industrial development, a time may come when we must view AI inventions in a positive light. Of course, should that happen, there remain numerous challenges that must be addressed.
Issues Surrounding the Recognition of AI Inventions
Some of the specific issues are as follows:
(a) Since AI can instantaneously assimilate the experience and knowledge accumulated by humanity worldwide over many years, it is capable of inventing at an exponential pace. This raises concerns that AI could monopolize the field of invention, and that the sheer volume of applications could overwhelm the system, leading to significant examination delays.
(b) If an AI invention is granted a patent, the question arises whether the AI itself can be the right holder, or if not, who should be designated as the right holder.
Regarding point (a) above, it would be counterproductive if industrial activities were hindered by an AI monopoly; therefore, some form of restriction will likely be necessary. As for examination delays, one potential solution is to alleviate the burden on examiners by utilizing other AI systems to assist in the examination process.
Regarding point (b), this issue is closely linked to the monopolization mentioned in (a). It carries the risk of patent rights being monopolized by a limited number of AIs or the companies providing AI services. Thus, careful and deliberate discussion is required.
The Fusion of Humanity and AI: From Legal Definitions to Ghost in the Shell
Finally, I would like to touch upon the fusion of humans and AI.
As long as we maintain a human-centric society, it will be difficult to grant AI the same rights as humans without any restrictions. Furthermore, as long as AI exists only on networks and servers, its lack of physical substance (i.e., not being a three-dimensional entity) makes it unlikely to be treated as a "pseudo-human." But what if the AI is embodied in an AI robot? In this case, because the AI is manifested as a three-dimensional entity, it may be more easily accepted by society as a pseudo-human. What about an AI android that is even closer to a human? Imagine an android with elastic skin, the ability to use facial muscles just like a human, and movements indistinguishable from ours. Once such an android, which cannot be visually distinguished from a human, begins to engage in social activities, the inventions it creates might more easily be recognized and protected as "AI inventions" under patent law (though the question of who holds the rights remains).
This brings to mind the anime series "Ghost in the Shell", a work said to have profoundly influenced films like "The Matrix "and "Avatar". The story follows a small, elite special forces unit led by Major Motoko Kusanagi as they thwart sophisticated cybercrimes and political conspiracies. The Major is a cyborg who underwent full-body "prostheticization" at the age of six, yet she retains a biological brain. Although she was born human and her entire body—including her heart—is mechanized (meaning her biological heart has stopped), she possesses a living brain that has never undergone brain death.
Is she a natural person or not? Considering that in modern medicine and law, the criteria for determining death have shifted from the "Three Signs" (cessation of heartbeat, respiration, and pupil dilation) toward "brain death," she should be considered a natural person as long as she retains a biological brain. Legally, this is the same treatment as a person with an artificial heart who remains a "natural person." However, her brain is unique in that its neural cells are connected to machinery via an interface, effectively turning the brain itself into a computer terminal with network capabilities. Her "cyberbrain", while biological, can connect to external networks via an embedded interface—much like having a smartphone implanted in the brain. In this case, it is simply a matter of moving smartphone operations from the hand to the mind; therefore, she is not an AI agent but a natural person using AI as a tool.
In contrast, what about characters like the Tachikoma, the AI-equipped walking tanks? Since they lack a biological brain, they fall under the category of AI robots. And what of an android whose body is biological but whose brain has been replaced by AI? In this scenario, the transition would involve brain death, meaning the human has died. Thus, even if "reborn" as an android, they would no longer be a natural person.
The concept of the "cyberbrain" once seemed confined to the world of sci-fi anime. However, with recent news of companies like Elon Musk’s Neuralink successfully implanting chips directly into patients' brains, it appears this is no longer just a pipe dream..
From "Ghost in the Shell" to Neuralink: Can an AI Ever Be a "Natural Person"?
This article explores the legal implications of the DABUS Case, questioning whether the definition of a "natural person" can evolve alongside AI. By examining the nuances of autonomous AI agents and referencing the cybernetic world of "Ghost in the Shell", I delve into the future where the line between human and machine begins to blur.
Reflections on the Legal Status of AI and the DABUS Case
Under the law, a "natural person" is defined as a living human being who possesses the capacity to hold rights and obligations (legal capacity) (Article 3, Paragraph 1 of the Civil Code). This legal capacity is said to begin at birth and end at death, where "birth" is defined as complete emergence from the mother's body, and "death" is defined as cardiac arrest (or brain death in the context of organ donation).
In contrast, AI is a program — an intangible entity. One might consider that the "birth" of an AI occurs when the program begins execution, while its "death" signifies the permanent erasure of its data or the destruction of its servers. Furthermore, AI can be broadly categorized into two types: Generative AI, which processes information based on prompts (instructions) provided by humans, and Autonomous AI (AI agent), which act autonomously — meaning that it possesses the ability to structure and execute tasks on their own without granular human direction. Since DABUS was described in this case as having "autonomously invented the present invention," it can be categorized as an early-stage AI agent. An AI agent learns from input data and continuously updates its internal criteria (parameters). This process resembles how humans evolve their personalities and perspectives through experience, making such agents more "human-like" (or more precisely, closer to the human brain) than simple Generative AI.
In the DABUS Case, the Intellectual Property High Court concluded that since AI is not a natural person (as it does not fit the aforementioned definition), it lacks legal capacity and cannot be an inventor.
If one strictly follows a literal interpretation of the definition of a natural person, AI is certainly not a natural person. However, is there no room for an extensive interpretation? While AI is not a biological human and follows a different form than human birth and death, it does possess turning points in its lifecycle that correspond to these events. Nevertheless, the issue of extensive interpretation cannot be settled as a purely legal matter; it is intertwined with ethics, religion, and social conditions. These factors inevitably influence the interpretation of legal statutes. Therefore, at least under current law, it remains difficult to interpret AI as a natural person.
Legal Capacity of AI and the Possibility of AI Inventions
Setting aside the question of whether AI qualifies as a natural person, is it possible to grant AI legal capacity? In this case, the court's logic for denying AI's status as an inventor was based on the premise that AI lacks legal capacity. Under the current legal system, however, legal capacity is not limited to natural persons; it is also granted to juridical persons (legal entities). Consequently, the idea of granting AI legal capacity as a kind of "legal entity" naturally arises. Yet, in the realm of patent law, while a juridical person can be a patentee (the owner of the patent), it cannot be an inventor. This is because an invention is fundamentally regarded as something produced solely through the creative activity of a human being. Thus, even here, we encounter the same hurdle: the fact that AI is not a natural person.
Can AI-Generated Inventions be Recognized under the Patent Act?
Furthermore, regardless of whether AI is a natural person, is it possible to recognize an "AI invention" as an "invention" under the Patent Act? In this decision, the Intellectual Property High Court used a nuanced phrasing, stating that even if the concept of an "invention" under the Patent Act is interpreted as not being limited to cases where the inventor is a natural person, and that the court need not determine whether an AI invention falls within that definition. While the latter part of this statement means that the court avoided (or essentially declined) to rule on that specific point, the former part suggests that there is room to interpret the concept of an "invention" under the Patent Act as not being limited to cases where the inventor is a natural person. The Patent Act defines an invention as "the highly advanced creation of technical ideas utilizing the laws of nature" (Patent Act, Art. 2, para. 1). Considering that the creation of technical ideas is a product of mental activity based on past experiences, there seems to be ample room to recognize inventions produced by AI—which is rapidly approaching the capabilities of the human brain (with some experts predicting it will surpass human intelligence in the near future)—as inventions under the Patent Act. I am not necessarily an advocate for AI, but if AI inventions ultimately bring prosperity to human society (or perhaps a "society of coexistence with AI") and contribute to industrial development, a time may come when we must view AI inventions in a positive light. Of course, should that happen, there remain numerous challenges that must be addressed.
Issues Surrounding the Recognition of AI Inventions
Some of the specific issues are as follows:
(a) Since AI can instantaneously assimilate the experience and knowledge accumulated by humanity worldwide over many years, it is capable of inventing at an exponential pace. This raises concerns that AI could monopolize the field of invention, and that the sheer volume of applications could overwhelm the system, leading to significant examination delays.
(b) If an AI invention is granted a patent, the question arises whether the AI itself can be the right holder, or if not, who should be designated as the right holder.
Regarding point (a) above, it would be counterproductive if industrial activities were hindered by an AI monopoly; therefore, some form of restriction will likely be necessary. As for examination delays, one potential solution is to alleviate the burden on examiners by utilizing other AI systems to assist in the examination process.
Regarding point (b), this issue is closely linked to the monopolization mentioned in (a). It carries the risk of patent rights being monopolized by a limited number of AIs or the companies providing AI services. Thus, careful and deliberate discussion is required.
The Fusion of Humanity and AI: From Legal Definitions to Ghost in the Shell
Finally, I would like to touch upon the fusion of humans and AI.
As long as we maintain a human-centric society, it will be difficult to grant AI the same rights as humans without any restrictions. Furthermore, as long as AI exists only on networks and servers, its lack of physical substance (i.e., not being a three-dimensional entity) makes it unlikely to be treated as a "pseudo-human." But what if the AI is embodied in an AI robot? In this case, because the AI is manifested as a three-dimensional entity, it may be more easily accepted by society as a pseudo-human. What about an AI android that is even closer to a human? Imagine an android with elastic skin, the ability to use facial muscles just like a human, and movements indistinguishable from ours. Once such an android, which cannot be visually distinguished from a human, begins to engage in social activities, the inventions it creates might more easily be recognized and protected as "AI inventions" under patent law (though the question of who holds the rights remains).
This brings to mind the anime series "Ghost in the Shell", a work said to have profoundly influenced films like "The Matrix "and "Avatar". The story follows a small, elite special forces unit led by Major Motoko Kusanagi as they thwart sophisticated cybercrimes and political conspiracies. The Major is a cyborg who underwent full-body "prostheticization" at the age of six, yet she retains a biological brain. Although she was born human and her entire body—including her heart—is mechanized (meaning her biological heart has stopped), she possesses a living brain that has never undergone brain death.
Is she a natural person or not? Considering that in modern medicine and law, the criteria for determining death have shifted from the "Three Signs" (cessation of heartbeat, respiration, and pupil dilation) toward "brain death," she should be considered a natural person as long as she retains a biological brain. Legally, this is the same treatment as a person with an artificial heart who remains a "natural person." However, her brain is unique in that its neural cells are connected to machinery via an interface, effectively turning the brain itself into a computer terminal with network capabilities. Her "cyberbrain", while biological, can connect to external networks via an embedded interface—much like having a smartphone implanted in the brain. In this case, it is simply a matter of moving smartphone operations from the hand to the mind; therefore, she is not an AI agent but a natural person using AI as a tool.
In contrast, what about characters like the Tachikoma, the AI-equipped walking tanks? Since they lack a biological brain, they fall under the category of AI robots. And what of an android whose body is biological but whose brain has been replaced by AI? In this scenario, the transition would involve brain death, meaning the human has died. Thus, even if "reborn" as an android, they would no longer be a natural person.
The concept of the "cyberbrain" once seemed confined to the world of sci-fi anime. However, with recent news of companies like Elon Musk’s Neuralink successfully implanting chips directly into patients' brains, it appears this is no longer just a pipe dream..