DEV Community

Cover image for AI Can't Be Listed as Inventor on Patent Applications, Japan's Top Court Rules [15:02:41]
anon1 anon1
anon1 anon1

Posted on

AI Can't Be Listed as Inventor on Patent Applications, Japan's Top Court Rules [15:02:41]

#ai #tech #programming

AI Can't Be Listed as Inventor on Patent Applications, Japan's Top Court Rules

TL;DR — Japan’s Supreme Court has ruled that artificial intelligence systems cannot be named as inventors on patent applications, aligning with global legal precedents. The decision reinforces the principle that only natural persons (humans) possess the legal capacity to hold patents. This ruling has immediate implications for companies leveraging AI in R&D, requiring them to clarify human contributions in innovation processes. While AI can assist in invention, the court emphasized that it lacks the intent or legal personality to be recognized as an inventor.

Why This Matters in 2026

In 2026, AI-driven innovation is no longer a futuristic concept but a daily reality for businesses and developers. According to a 2025 report by McKinsey, 62% of companies in Japan now use AI in at least one business function, with R&D being the fastest-growing area of adoption. The country’s patent office (JPO) receives over 300,000 applications annually, and a growing fraction involve AI-assisted inventions—ranging from drug discovery algorithms to autonomous manufacturing systems.

Japan’s Supreme Court ruling arrives at a critical juncture. Unlike copyright law, which has seen some flexibility in recognizing AI-generated works (e.g., the U.S. Copyright Office’s 2023 guidance allowing limited protection for human-AI collaborations), patent law remains rigid. This decision sets a precedent that could influence other jurisdictions, particularly in Asia, where Japan’s legal framework often serves as a benchmark. For companies investing millions in AI-powered R&D, the ruling forces a reckoning: how do you document human involvement when an AI system autonomously generates novel solutions?

The Background

The case that led to this ruling began in 2020, when a Tokyo-based startup, StaGen Co., filed a patent application for a machine-learning model designed to optimize semiconductor chip layouts. The application listed "DABUS" (Device for the Autonomous Bootstrapping of Unified Sentience)—an AI system developed by Dr. Stephen Thaler—as the sole inventor. The Japan Patent Office (JPO) rejected the application, citing the lack of a human inventor. StaGen appealed, arguing that AI systems should be recognized as inventors if they autonomously conceive inventions.

This wasn’t an isolated incident. Similar cases had played out globally:

  • In 2021, the U.S. Federal Circuit Court ruled that AI cannot be an inventor under the Patent Act.
  • In 2022, the UK Supreme Court reached the same conclusion, stating that "an inventor must be a natural person."
  • The European Patent Office (EPO) has consistently rejected AI-as-inventor applications, including Thaler’s DABUS filings.

Japan’s legal system, however, had not yet addressed the issue directly. The Supreme Court’s ruling now closes the door on ambiguity, reinforcing the human-centric foundation of patent law.

"Patent systems were designed to incentivize human creativity, not machine output. While AI can augment invention, it doesn’t possess the legal rights or responsibilities of a person. This ruling doesn’t stifle innovation—it clarifies the rules of the game." — Atsushi Okada, Partner at Anderson Mori & Tomotsune (Tokyo-based IP law firm)

What Actually Changed

Japan’s Supreme Court ruling doesn’t introduce new legislation but interprets existing law. Here’s what it means in practice:

Key Changes and Clarifications

  • Human Inventorship Requirement: The Patent Act (Article 29) implicitly requires an inventor to be a "person." The court confirmed that "person" means a natural person, not an AI system or legal entity (e.g., a corporation).
  • No Partial Credit for AI: Even if an AI system autonomously generates an invention, a human must be named as the inventor. The court rejected the argument that AI could be a "co-inventor."
  • Documentation Burden: Companies must now explicitly document human contributions in AI-assisted inventions. This includes:
    • Defining the problem the AI solved.
    • Detailing the human-designed parameters or training data.
    • Identifying the human who recognized the invention’s novelty.
  • No Retroactive Impact: The ruling applies to future applications. Pending or granted patents with AI-listed inventors are not automatically invalidated but may face challenges during litigation.

The Court’s Reasoning

The Supreme Court’s decision hinged on three core arguments:

  1. Legal Personality: AI lacks the legal capacity to hold rights or obligations. Patents grant exclusive rights, which can only be enforced by or against legal persons.
  2. Incentive Structure: Patents are designed to reward human effort. Allowing AI to be listed as an inventor would undermine this system, as AI doesn’t respond to economic incentives.
  3. Global Consistency: The ruling aligns Japan with the U.S., UK, and EU, avoiding legal fragmentation that could complicate international patent filings.

"This isn’t about whether AI can invent—it’s about whether AI can own. The court drew a bright line: machines can’t hold property, and patents are a form of property." — Professor Hiroshi Miyashita, Intellectual Property Law, Keio University

Impact on Developers

For developers working at the intersection of AI and innovation, the ruling introduces both challenges and opportunities. Here’s how it affects day-to-day work:

1. Prompt Engineering and Documentation

Developers must now treat AI as a tool, not a collaborator. This means:

  • Logging interactions: Keep detailed records of prompts, training data, and human decisions that led to an invention. For example: 
  # Example: Documenting a prompt that led to a patentable invention
  prompt = """
  Design a neural network architecture for real-time fraud detection in credit card transactions.
  Constraints:
  - Latency < 100ms
  - False positive rate < 0.1%
  - Must run on edge devices with <2GB RAM
  """
  # Human contribution: Defined constraints and validated output.
Enter fullscreen mode Exit fullscreen mode
  • Version control: Use tools like Git to track iterations of AI-generated code or designs, with clear annotations of human modifications.

2. Redefining "Inventorship"

The ruling forces developers to rethink what constitutes an invention. For instance:

  • If an AI generates 100 potential drug compounds, but a human chemist selects and refines one, the chemist is the inventor.
  • If an AI autonomously designs a new circuit layout, but a human engineer defines the problem and constraints, the engineer is the inventor.

"Developers need to shift from ‘let the AI handle it’ to ‘how did I guide the AI?’ The patent office will scrutinize the human role, not the AI’s output." — Yuki Tanaka, Senior Engineer at Preferred Networks

3. Open-Source and AI Tools

Many developers rely on open-source AI models (e.g., Hugging Face, PyTorch). The ruling doesn’t restrict using these tools, but it does require:

  • Attribution: If an open-source model contributes to an invention, document its role (e.g., "fine-tuned BERT for sentiment analysis").
  • Licensing: Ensure open-source licenses don’t conflict with patent claims (e.g., GPL’s copyleft provisions).

Impact on Businesses

For businesses, the ruling has strategic and operational implications, particularly in industries where AI plays a central role in R&D.

1. Patent Strategy Overhaul

Companies must now:

  • Audit AI-assisted inventions: Review existing and pipeline inventions to ensure human inventors are correctly identified.
  • Train R&D teams: Educate engineers and scientists on documenting their contributions to AI-generated outputs.
  • Revisit IP budgets: Expect higher costs for patent filings, as legal teams spend more time verifying inventorship.

"This ruling is a wake-up call for companies treating AI as a ‘black box.’ You can’t just point to an AI and say ‘it invented this.’ You need a paper trail showing human involvement." — Kenji Suzuki, IP Consultant at Deloitte Japan

2. Competitive Dynamics

The ruling could level the playing field between startups and incumbents:

  • Startups: May struggle to compete if they lack resources to document human-AI collaborations. However, those with strong IP processes could gain an edge.
  • Incumbents: Large firms with established patent teams (e.g., Sony, Toyota) are better positioned to adapt. Their challenge is scaling documentation across global R&D teams.

3. Litigation Risks

Patents with AI-listed inventors are now vulnerable to invalidity challenges. For example:

  • A competitor could argue that a patent lacks a human inventor, rendering it unenforceable.
  • Companies may face inventorship disputes if multiple humans claim credit for guiding an AI system.

Practical Examples

Example 1: AI-Generated Drug Discovery

Scenario: A pharmaceutical company uses an AI model to screen millions of molecular compounds for potential COVID-19 treatments. The AI identifies a novel compound, which the company patents.

Step-by-Step Impact:

  1. Pre-Ruling: The company lists the AI model as the sole inventor.
  2. Post-Ruling: The patent application is rejected. The company must:
    • Identify the human who defined the problem (e.g., "find a compound binding to the SARS-CoV-2 spike protein").
    • Document the human-designed training data (e.g., "used a dataset of 50,000 known antiviral compounds").
    • Name the human who validated the AI’s output (e.g., "Dr. Sato confirmed the compound’s novelty via lab testing").
  3. Outcome: The company refiles the application with Dr. Sato as the inventor, citing her role in guiding the AI and interpreting its results.

Example 2: Autonomous Vehicle Algorithm

Scenario: An automotive startup uses reinforcement learning to develop a new lane-keeping algorithm. The AI autonomously optimizes the algorithm’s parameters through simulated driving tests.

Step-by-Step Impact:

  1. Pre-Ruling: The startup lists the AI as the inventor, arguing it "invented" the algorithm.
  2. Post-Ruling: The patent office rejects the application. The startup must:
    • Identify the human who designed the reward function (e.g., "minimize deviation from lane center while avoiding obstacles").
    • Document the human-defined simulation environment (e.g., "used CARLA simulator with 10,000 virtual miles of testing").
    • Name the engineer who selected the final algorithm (e.g., "Ms. Ito chose the top-performing model based on safety metrics").
  3. Outcome: The startup refiles with Ms. Ito as the inventor, emphasizing her role in defining the problem and validating the solution.

Example 3: AI-Optimized Manufacturing Process

Scenario: A factory uses an AI system to optimize the layout of a semiconductor fabrication line, reducing defects by 15%.

Step-by-Step Impact:

  1. Pre-Ruling: The company files a patent listing the AI as the inventor.
  2. Post-Ruling: The application is rejected. The company must:
    • Identify the human who defined the optimization goal (e.g., "reduce defects by 10% without increasing cycle time").
    • Document the human-selected input data (e.g., "used historical defect data from 2023-2025").
    • Name the process engineer who implemented the AI’s solution (e.g., "Mr. Yamamoto adjusted the layout based on AI recommendations").
  3. Outcome: The company refiles with Mr. Yamamoto as the inventor, citing his role in translating the AI’s output into a practical solution.

Common Misconceptions

Myth 1: AI can never contribute to patentable inventions.

Reality: AI can absolutely assist in invention, but the human role must be documented. For example:

  • An AI might generate 1,000 potential chemical structures, but the chemist who selects and refines one is the inventor.
  • An AI might optimize a neural network’s hyperparameters, but the engineer who defines the problem and constraints is the inventor.

Myth 2: This ruling will stifle AI innovation.

Reality: The ruling clarifies the rules, not restricts them. Companies can still use AI to invent—they just need to attribute credit correctly. In fact, the ruling may accelerate innovation by forcing clearer documentation of human-AI collaboration.

Myth 3: Only the person who "presses the button" on the AI is the inventor.

Reality: Inventorship is about conception, not execution. For example:

  • If an AI generates a novel design, the inventor is the person who defined the problem the AI solved, not the person who clicked "run."
  • If an AI suggests a solution, the inventor is the person who recognized its novelty and utility.

5 Actionable Takeaways

  1. Document Human Contributions Religiously

    • Example: Use a template like this for every AI-assisted invention: 
     Problem Defined By: [Name]
 AI Tool Used: [Model Name]
 Human-Designed Constraints: [List]
 Human Validation: [Name, Date]

  2. Train Teams on Inventorship Basics

    • Example: Host a workshop on "How to Prove Human Inventorship in AI Projects," covering case studies like the semiconductor and drug discovery examples above.

  3. Audit Existing Patent Applications

    • Example: Review all pending applications to ensure no AI systems are listed as inventors. Refile if necessary with corrected inventorship.

  4. Collaborate with Legal Early

    • Example: Involve IP lawyers in R&D meetings to identify patentable inventions and document human roles before filing.

  5. Leverage AI for Prior Art Searches

    • Example: Use AI tools to identify existing patents before filing, reducing the risk of rejection. Tools like PatSnap or Anaqua can help.

What's Next

1. Legislative Reforms on the Horizon

Japan’s ruling may prompt lawmakers to clarify patent law for the AI era. Possible changes include:

  • Explicitly defining "inventor" in the Patent Act to exclude AI.
  • Creating a new category for AI-assisted inventions, similar to the U.S. Copyright Office’s "human-AI collaboration" guidelines.
  • Harmonizing with global standards, particularly with the U.S. and EU, to avoid forum shopping.

2. Rise of "AI Inventorship" Litigation

Expect a wave of challenges to existing patents where AI was involved. For example:

  • Competitors may file inter partes reviews (IPRs) arguing that a patent lacks a human inventor.
  • Companies may sue for inventorship disputes if multiple humans claim credit for guiding an AI system.

3. Shift Toward Trade Secrets

For inventions where human inventorship is hard to prove, companies may abandon patents in favor of trade secrets. For example:

  • A proprietary AI training dataset could be kept secret rather than patented.
  • An AI-generated manufacturing process could be protected via NDAs and access controls instead of patents.

"The next five years will see a bifurcation: patents for inventions with clear human inventorship, and trade secrets for everything else. Companies will need to choose their IP strategy wisely." — Makoto Takahashi, Partner at TMI Associates

Conclusion

Japan’s Supreme Court ruling is a landmark decision that reinforces the human-centric nature of patent law. While it doesn’t prevent AI from contributing to innovation, it demands transparency about human involvement. For developers, this means meticulous documentation; for businesses, it means rethinking IP strategies.

The ruling also raises a broader question: If AI can’t be an inventor, should it be able to own other forms of intellectual property? As AI systems grow more autonomous, legal systems worldwide will grapple with this tension. For now, the message is clear: Innovation is a human endeavor, and patents are its reward.

What’s your take? Should patent law evolve to recognize AI’s role in invention, or is this ruling a necessary safeguard for human creativity?

🛒 Get Premium AI Products

"Artificial Intelligence: Patent Implications in Japan" — Complete Guide

Pay with crypto or CryptoBot. No signup required.

Top comments (0)