Intellectual Property Rights Made Easy: Leveraging the System Wisely in the Era of AI Agents

Intellectual Property Rights Even a Rabbit Can Understand — Leveraging the System Wisely in the AI Agent Era —

Introduction

Hello everyone! Welcome to the "Even a Rabbit Can Understand" series, pyon! In this article, we'll explain Intellectual Property Rights, such as patents and trademarks, with a focus on the AI agent era.

"Patents," "utility models," "designs," "trademarks"... You might have heard these words, but many of you might not know exactly how they differ or how to utilize them effectively. Especially in recent years, with the spread of AI agents, the importance of protecting and utilizing intellectual property has become increasingly high.

In this article, we will explain everything from the basic concepts of intellectual property rights to specific utilization methods in the AI agent era as clearly as possible. This content is designed to answer questions for engineers and product developers, such as "How should I protect my company's technology and services?" and "What kind of rights should I acquire?"

Basics of Intellectual Property Rights

First, let's understand what intellectual property rights are. Intellectual property rights are rights that protect things created through human intellectual creative activity. Among them, we will look closely at four types of rights called industrial property rights (patent rights, utility model rights, design rights, and trademark rights).

Patent Rights: Protecting Technical Ideas

Patent rights are rights that protect inventions. Here, "invention" refers to highly advanced creations of technical ideas utilizing the laws of nature. Examples include new machines, devices, manufacturing methods, and chemical compositions.

The duration of a patent right is relatively long, lasting 20 years from the filing date (certain pharmaceuticals can be extended by up to 5 years), and it offers the strongest protection. However, a characteristic feature is that the examination process takes time and costs money.

In AI-related inventions, new algorithms, data processing methods, and business models using AI can be subject to patents.

Utility Model Rights: Protecting Small Improvements Quickly and Cheaply

Utility model rights are rights that protect ideas related to the shape, structure, or combination of articles. Even if they are not as advanced in technical ideas as patents, they can protect "small improvements."

The biggest feature is that they are registered without examination, allowing for the acquisition of rights in about three months from application. The duration is 10 years from the filing date, which is shorter than a patent, but the costs are also lower. However, a request for a technical opinion report is required when exercising the rights.

They are suitable for improvements that are not as technically advanced as patents, such as improvements in the shape or structure of AI hardware.

Design Rights: Protecting Product Design

Design rights are rights that protect the design of an article (shape, pattern, color, etc.). They cover things that evoke an aesthetic sense through vision.

The duration is 25 years from the filing date, and they are registered in a relatively short period (approximately 6 months). They are suitable for protecting the appearance design of AI devices or the design of user interfaces.

Trademark Rights: Protecting the Brand

Trademark rights are rights that protect the "brand" of a product or service. They cover signs used to distinguish one's own products or services from those of other companies, such as names, logos, and slogans.

The duration is 10 years from the registration date, but a key feature is that the rights can be maintained semi-permanently through renewal procedures. They are ideal for protecting brand identity, such as AI service names and logos.

Intellectual Property Rights Acquisition Process

Let's look at the basic flow of procedures for acquiring intellectual property rights.

Flow from Application to Registration

There are similarities and differences in the acquisition process for each right.

For Patents:

1. Preparation of application documents
2. Patent application
3. Request for examination (within 3 years from the filing date)
4. Substantive examination (examination of novelty, inventive step, etc.)
5. Decision to grant a patent (or decision of rejection -> appeal possible)
6. Patent registration (payment of patent fees)

For Utility Models:

1. Preparation of application documents
2. Application for utility model registration
3. Basic requirement examination (formality requirements only)
4. Utility model registration (payment of registration fees)
5. (Before exercising rights) Request for technical opinion report

For Designs:

1. Preparation of application documents
2. Application for design registration
3. Substantive examination (examination of novelty, creativity, etc.)
4. Decision of registration (or decision of rejection -> appeal possible)
5. Design registration (payment of registration fees)

For Trademarks:

1. Preparation of application documents
2. Application for trademark registration
3. Substantive examination (examination of distinctiveness, similarity, etc.)
4. Decision of registration (or decision of rejection -> appeal possible)
5. Trademark registration (payment of registration fees)
6. (Every 10 years) Renewal procedures

Comparison of Costs and Periods for Each Right

There are significant differences in the costs and periods required to acquire each right.

• Patent: Approximately 600,000 to 800,000 yen, around 4 years from application to registration
• Utility Model: Approximately 250,000 to 300,000 yen, around 3 months from application to registration
• Design: Approximately 200,000 to 250,000 yen, around 6 months from application to registration
• Trademark: Approximately 100,000 to 150,000 yen, around 8 to 12 months from application to registration

Please note that these are general estimates and may vary depending on the complexity of the case or the examination status. Also, if you are considering acquiring rights overseas, be aware that additional costs will be incurred.

Which Right Should You Choose? Points for Judgment

You may often be unsure which right to acquire within a limited budget. Please refer to the following points:

• Patents for technically innovative things: Patents are suitable for new technologies or algorithms that directly lead to differentiation from competitors.
• Utility models for small improvements: Recommended for small innovations where you want to acquire rights quickly or are not confident in the patentability.
• Designs for external appearance: Protect the look of a product or the design of operation screens with design rights.
• Trademarks for service names and logos: Trademarks are essential for long-term brand building.

In practice, it is ideal to combine these rights for multilayered protection. For example, for an AI service, you can achieve stronger protection by combining multiple rights, such as using a patent for the core technology, a design right for the UI design, and a trademark for the service name.

Intellectual Property Strategy in the AI Era

With the rapid development of AI technology, intellectual property strategies are entering a new phase. Let's look at intellectual property strategies that are particularly important in the AI agent era.

Acquiring Rights for AI-Generated Creations

Discussions regarding the copyright of AI-generated content (text, images, music, etc.) are becoming more active. Under Japanese copyright law, a work is defined as "a creative expression of thoughts or emotions," and it is generally considered that something autonomously generated by AI does not fall under the category of a "work."

However, if a human is creatively involved in the AI generation process (e.g., through prompt engineering, selection, or modification of the output), it may be protected as a human work. The key is to clarify the elements of human creative involvement when utilizing AI tools.

Furthermore, designs or logos generated by AI can be filed for design rights or trademark rights. In such cases, there is no need to explicitly state that it is an "AI-generated product," and the application can be filed as a regular design or trademark.

Intellectual Property Management Utilizing AI

AI can also be used for the management of intellectual property itself. For example, it can be used in the following ways:

• Streamlining prior art searches: Using AI to extract relevant prior art from vast amounts of patent information.
• Automating monitoring for rights infringement: Monitoring with AI to see if competitors' products or services infringe on your company's rights.
• Optimizing IP portfolios: Visualizing your company's strengths and weaknesses through AI-based analysis and formulating strategic filing plans.

In fact, the Japan Patent Office (JPO) is also proceeding with the introduction of AI-based patent examination support systems, and the relationship between AI and intellectual property is deepening further.

Strategy with an Eye on Global Expansion

Since AI services can easily cross borders via the internet, a global intellectual property strategy is essential.

• PCT Application (Patents): A system that allows for simultaneous applications in multiple countries with a single application procedure.
• Madrid Protocol (Trademarks): A system that allows for efficient trademark registration in multiple countries through an international application.
• Hague Agreement (Designs): A system that enables the acquisition of design rights in multiple countries with a single international application.

By utilizing these international application systems, you can build a cost-effective global IP strategy. Acquiring rights in countries with significant AI technology development, such as China and the United States, is particularly important.

Case Studies: AI and Intellectual Property

Patent Examples Related to Generative AI

Patent applications related to generative AI technology have increased rapidly in recent years. According to the Japan Patent Office's "Cases of Patent Examination for AI-related Technology," 10 new patent examination cases involving generative AI were added as of March 13, 2024.

For example, in patent cases related to image-generating AI technology, things like "methods for generating another image based on a specific input image" or "preprocessing techniques for generating high-quality images from text prompts" are subjects of examination.

Additionally, methods for making AI models lightweight and techniques for streamlining the training of AI models have also been patented. These are becoming important technologies for the development of edge devices that incorporate generative AI.

Patent Strategy for Services Utilizing AI

For services utilizing AI, a strategy of patenting the service delivery method itself as a "business-related invention" is effective.

For example, medical diagnosis support systems using AI or real estate price prediction systems using AI can be patented as methods for solving industry-specific issues with AI. Such patents become powerful weapons for protecting business models.

The key to acquiring a patent for an AI service is to clarify the technical issue and specifically describe the technical means for how AI solves it. Note that simply "using AI" is not enough for patentability to be recognized.

Points to Learn from Success Stories and Failures

From success stories of AI startups, the importance of building an early IP strategy can be seen. For example, a company that grew rapidly with image recognition AI technology strategically performed design registration for its user interface and trademark registration for its service name at the same time as acquiring patents for its core technology, thereby reducing the risk of imitation from competitors.

On the other hand, there are many examples where a delay in IP strategy hindered a company's growth. There are cases where startups that focused so much on developing AI technology that they put IP protection on the back burner were restricted in their business expansion because competitors acquired patents first.

The point to learn from these cases is the importance of proceeding with an IP strategy in parallel with product development. Especially in the field of AI agents, where the speed of technological innovation is fast, early acquisition of rights directly leads to securing a competitive advantage.

Points for Wisely Utilizing Intellectual Property Rights

Finally, let's summarize wise ways to use intellectual property rights in the AI agent era.

Open Strategy and Closed Strategy

Intellectual property strategies can be broadly divided into "open strategies" and "closed strategies."

Open strategy is a strategy to expand the market by actively disclosing technology and know-how and licensing them widely. For example, a model that releases basic AI algorithms as open source and generates revenue from applications and services built on top of them fits this description.

Closed strategy is a strategy to maintain exclusive rights by protecting core technologies as patents or trade secrets. It is effective when you have unique technology that clearly differentiates you from competitors.

Many companies adopt a "selective open & closed strategy" that combines both. For example, they might keep basic API specifications open while protecting the underlying advanced AI algorithms as closed.

Deciding between Patenting and Secrecy

Not all technologies and know-how should be turned into rights such as patents. In some cases, it may be more advantageous to keep them as "trade secrets."

Suitable for rights acquisition:

• Things where the technical content can be inferred by looking at the product or service
• Things that competitors are highly likely to develop independently
• Things where license income can be expected

Suitable for secrecy:

• Things that are difficult to reverse-engineer (e.g., AI training methods)
• Know-how whose details would be disclosed if a patent application is filed
• Things that require protection for a longer period than what a patent offers

In AI agent development, for example, it is effective to use different approaches: keeping data preprocessing methods or model tuning methods as trade secrets, while patenting or registering designs for UIs and implementation algorithms.

Preparing for the AI Agent Era

The AI agent era is also changing the nature of intellectual property itself. Here are some points to prepare for the future:

1. Regular IP Inventory: Conduct regular inventories of your company's intellectual property and re-evaluate its value in the AI agent era.

2. Acquisition of Rights Assuming AI Utilization: Consider filing patent applications with a view toward future AI integration or selecting trademark classifications that anticipate use cases in the AI era.

3. Review of Contracts: Check whether existing IP-related contracts (such as license agreements) will be affected by the emergence of AI agents, and update them as necessary.

4. Balancing AI Ethics and IP: When using AI agents, strive for appropriate use not only from the perspective of protecting intellectual property rights but also from the perspective of AI ethics.

Intellectual property strategy, like technical development, requires continuous review and improvement. It is important to update your strategy flexibly and rapidly in line with the rapid development of AI.

Summary

In this article, we explained the basics of the four types of intellectual property rights—patents, utility models, designs, and trademarks—and how to utilize them in the AI agent era.

Intellectual property rights are an essential system for promoting innovation and protecting the rights of creators. Especially with the spread of AI agents, as new ways of creation and invention emerge, their importance is increasing more than ever.

I hope that engineers and product developers will keep an eye on the creation and protection of intellectual property alongside their daily development activities. Having perspectives such as "how to protect my own ingenuity" and "whether I am infringing on the rights of others" is the key to sustainable innovation.

In the "Even a Rabbit Can Understand" series, we will continue to deliver useful knowledge for the AI era in an easy-to-understand way, pyon!

Reference Materials

• Ministry of Economy, Trade and Industry, Japan Patent Office "Clearly Understanding Intellectual Property Rights"
• Japan Patent Office "About Patent Examination Cases for AI-related Technology"
• Cabinet Office "Intellectual Property Rights Study Group in the AI Era"
• Intellectual Property Promotion Plan 2023 (Decision by the Intellectual Property Strategy Headquarters on June 9, 2023)