Introduction
Antonio Campinos, President of the European Patent Office, commonly known as the EPO, responded to written questions from Kyodo News and pointed out that the increase in inventions related to artificial intelligence, or AI, is “raising new challenges” for the patent system. Patent applications for AI-related technologies are increasing rapidly worldwide, and their subject matter now extends across a wide range of fields, including medicine, energy, and transportation. Another notable feature is that the actors filing such applications are no longer limited to major corporations, but now also include startups, research institutions, and other entities, meaning that the main drivers of technological development are becoming increasingly multipolar. Against this background, the fundamental issue at the core of the patent system—that “the inventor must be a human being”—has once again come to the fore in relation to inventions involving AI.
The Increase in AI-Related Inventions Is Not Merely a Technological Trend
What is important about the recent remarks is that the increase in AI-related inventions is not being treated simply as a rise in the number of patent applications. AI is not a technology used only in certain specific industries. It is becoming a foundational technology for almost every industry, including medical diagnosis, drug discovery, power control, autonomous driving, logistics, manufacturing, finance, and content generation.
In other words, the increase in AI-related patents is not merely a matter for the information and communications field. AI is cutting across areas of technological development that have traditionally been organized by field, such as machinery, chemistry, electricity, medicine, architecture, and transportation. Until now, the patent system has developed mechanisms for examining inventions by technical field and evaluating their differences from prior art. However, as AI becomes embedded in diverse fields, it will become increasingly important in examination practice to distinguish between the “technical features of AI itself” and the “technical features of the applied field in which AI is used.”
The Core Issue Is Whether AI Invented It or a Human Invented It
The issues surrounding AI-related inventions can broadly be divided into two aspects. The first concerns inventions relating to AI itself. Examples include new learning models, inference methods, data processing methods, image recognition technologies, and natural language processing technologies.
The second concerns inventions created using AI as a tool. For example, AI may propose candidate compounds, and a human may select promising substances from among them and identify pharmaceutical uses. Or AI may generate a large number of design proposals, and a human may select the optimal structure and commercialize it as a product.
The former can be organized relatively easily as “inventions relating to AI.” The latter, however, involves the fundamental question of “who is the inventor.” Did AI merely propose candidates? Did AI play a role that was substantially close to completing the invention? Was the human merely a person who confirmed the result? Or was the human the entity that embodied the technical idea? This boundary will become increasingly ambiguous in the future.
The Meaning of the Principle That “the Inventor Must Be a Human Being”
The issue identified by President Campinos—that “the inventor must be a human being”—is not a mere matter of formality. The patent system grants exclusive rights to the person who made the invention, or to that person’s successor in title. In that system, the concept of the inventor is the starting point of rights.
If AI were recognized as an inventor, a series of questions would arise: Does AI have legal capacity? To whom would rights be transferred from AI? Would the owner of the AI become the rights holder? Would the user who used the AI become the rights holder? This is not simply a question of what name should be written in the inventor field of a patent application. It concerns the very ownership of the right to obtain a patent.
On the other hand, even if AI is not recognized as an inventor, problems remain. For inventions in which AI plays an extremely large role, it will be necessary to clarify which human acts should be evaluated as “inventing.” Is a person an inventor merely by entering a prompt? Is the inventor the person who evaluates the AI output and identifies its technical significance? Is it the person who confirms the effect through experiments? In practice, this determination will become very difficult.
In Corporate Practice, “Records of the Invention” Will Become More Important
What will become important in patent practice in the AI era is how concretely the process leading to completion of the invention can be recorded. In research and development using AI, it will be necessary to keep records of who set what kind of problem, which AI tool was used, what inputs were made, which outputs were adopted, and what technical judgments were added.
This point is particularly important in joint research and collaborations with startups. When AI tool providers, data providers, researchers, development personnel, experimental personnel, and business companies are all involved, it can easily become unclear who the inventors are and to whom the rights belong.
Going forward, for inventions involving AI, organizing research notes, prompt histories, output results, adoption or rejection decisions, experimental results, and improvement processes will become an important management item before filing a patent application. This does not mean that patents cannot be obtained simply because AI was used. However, it will become important to be able to explain how AI was used and where humans were involved in technical creation.
As Multipolarization Advances, Adjusting the Patent System Will Become Even More Difficult
The “multipolarization” pointed out by President Campinos should also not be overlooked. AI-related inventions are being created not only by major corporations, but also by startups, universities, research institutions, and individual developers. In the sense that the base of innovation is expanding, this is a welcome development.
However, as the actors become more diverse, the operation of the patent system also becomes more complex. Major corporations often have systems in place for invention disclosures, employee invention regulations, contracts, and checks by intellectual property departments. By contrast, startups and research institutions may not have sufficiently organized rules regarding the terms of use of AI tools, joint research agreements, rights in data, and inventor determination.
As a result, even promising AI-related inventions may later become a source of disputes if the ownership of rights and the identification of inventors are insufficiently organized at the time of filing. In the AI era, not only the speed of technological development but also the ability to design intellectual property management systems will become part of competitiveness.
Why International Coordination Is Necessary
The issue of inventorship in AI-related inventions cannot be solved by one country alone. Patents are governed by national systems, but AI-related businesses are developed across borders from the outset. If Europe, Japan, the United States, China, South Korea, and other jurisdictions differ significantly in their views on inventorship and their treatment of AI-generated inventions, it will become necessary to adjust filing strategies and rights ownership structures country by country for the same technology.
In this sense, it is significant that the EPO President expressed an intention to work in coordination with relevant authorities in other countries. Even if it is not necessary to create completely identical systems, there is a need to form an international common understanding, at least regarding what kind of human involvement is required for inventorship to be recognized in inventions involving AI, and what kinds of descriptions and evidence will be necessary.
AI technology evolves rapidly, while the patent system is a system that emphasizes the stability of rights. How to bridge this difference in speed will be a major challenge going forward.
The Patent System in the AI Era Will Redefine “Human Creativity”
The more deeply AI becomes involved in the process of invention, the more the patent system will be forced to reconsider what it means for “humans to invent.” Traditionally, it was taken for granted that humans would be responsible for the entire process of identifying a problem, considering means for solving it, and confirming its effects. However, in an era where AI generates candidates, optimizes designs, and proposes experimental conditions, the role of humans will change.
What will become important is not whether humans thought of everything entirely on their own. Rather, the key question will be whether humans technically evaluated AI outputs, gave them meaning in relation to the problem to be solved, embodied them in an implementable form, and completed them as technologies useful to society.
In other words, inventorship in the AI era is likely to shift from the simple question of “who first came up with the idea” to the question of “which human beings made an essential contribution to the technical creation.”
Conclusion
The remarks by the EPO President are not merely an expression of general caution regarding the increase in AI-related patents. They encompass multiple issues at the core of the patent system, including the concept of the inventor, ownership of rights, examination practice, and international coordination.
AI is a powerful tool that assists invention. However, the more powerful AI becomes, the more important it becomes to evaluate human creative involvement. In future patent practice, what will be required is not simply whether AI was used, but a clear explanation of what technical judgments humans made through the use of AI.
The patent system in the AI era cannot be organized simply as a binary choice between recognizing AI as an inventor or not. Rather, the real question is how to reconcile human creativity, AI assistance, technical contribution, and the stability of rights. The EPO President’s remarks indicate that this debate is now entering a full-fledged stage.