On July 17, 2024, the United States Patent and Trademark Office (USPTO) issued guidance addressing patent eligibility for AI-related inventions. This move aims to clarify how AI inventions align with patent subject matter eligibility rules, in line with Executive Order 14110, which promotes the “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.” The new guidance addresses key challenges faced by innovators and provides a structured approach for evaluating AI inventions.
This blog provides a comprehensive overview of the USPTO’s updated AI technology patent guidance, its implications for innovators, and practical strategies to navigate the patenting process.
USPTO’s Role in AI and Emerging Technologies
The USPTO plays a critical role in the patent process for AI inventions. Patent systems must evolve to protect technological innovation while ensuring that existing legal frameworks remain relevant. The guidance on AI technologies encourages AI development that is both secure and ethical.
Key Highlights of the 2024 Guidance Update
Streamlining AI Patent Eligibility
The 2024 guidance update on patent subject matter eligibility, including AI, ensures that only innovative AI solutions qualify for patent protection. This helps inventors clarify what qualifies as patent-eligible subject matter. While the Alice/Mayo framework for assessing subject matter eligibility remains unchanged, the updated guidance provides valuable insights as AI technology rapidly evolves. The USPTO maintains its commitment to granting patents for AI inventions, including those utilising AI. Nevertheless, it is crucial to demonstrate a technical solution to a technical problem.
Patent Eligibility Process
The updated guidance provides a two-step analysis for determining AI invention eligibility.
- Step 2A – Prong 1
The first inquiry (Step 2A, Prong 1) asks whether a claim recites a judicial exception that warrants further analysis. The USPTO examiner must assess whether a claim encompasses features that are classified as abstract ideas. These abstract ideas typically involve mathematical concepts, specific methods of organising human activity, or mental processes.
The 2024 Updated Guidance enumerates a list of examples that provide clarity on what is considered an abstract idea.
Physical Implementations
The Guidance clarifies that claims directed to physical implementations of abstract ideas, such as ASICs for artificial neural networks or systems for monitoring livestock health, are not considered as abstract ideas. These claims are eligible for patent protection because they involve tangible elements and specific applications.
Practical Applications
Claims that involve mathematical concepts but are applied to specific, practical problems, such as classifying and sorting particles in real-time using flow cytometry, are also not considered abstract ideas. These claims are eligible as long as they do not recite a mathematical concept or other abstract idea under Step 2A, Prong One.
Managing Personal Behaviour or Relationships
Claims directed to methods of collecting and recording information about a user’s movements and location history, such as creating a digital travel log, are considered as abstract ideas as they involve managing personal behaviour. This is evident in the case of Weisner v. Google LLC.
- Step 2A – Prong 2:
The second inquiry (Step 2A, Prong 2) asks whether the claim integrates the alleged judicial exception into a practical application. Examiners must assess whether a claim, in its entirety, integrates the purported abstract idea into a practical application. If the AI invention is applied to solve real-world problems or enhance functionality, it may pass this test. The guidance further clarifies that claim elements should “apply, rely upon, or utilise the judicial exception in a way that imposes a meaningful limitation on the judicial exception.”
In the domain of AI technologies, patent applicants frequently fulfil the requirements of Step 2A, Prong 2, by demonstrating that their AI invention enhances the performance of the computer or advances another technology or technical field. This is often referred to as a “technical solution to a technical problem.”
The Guidance recognizes that numerous AI invention claims are eligible as improvements to computer or technological functions. It aims to elucidate the distinction between eligible and ineligible claims in this context. The guidance clarifies that eligible claims must substantiate a specific improvement as detailed in the patent specification.
The guidance provides examples of claims that sufficiently describe “improvements” and therefore satisfy Step 2A, Prong 2:
Mathematical Process vs. Technological Improvement
Claims directed to a method of resolving a haplotype phase using a Hidden Markov Model were deemed ineligible because they primarily involved an improved mathematical process rather than a technological improvement. The court determined that the claims did not provide a specific improvement to a technological process.
The Hidden Morov model refers to how patent law assesses the eligibility of innovations involving computational models. It helps estimate the probability of a technology being at a certain stage of its life cycle and identify patterns1.
Specific Functional Rules for Computer Animation
In contrast, claims for a rule-based system to animate lip synchronisation were found eligible. This was because the claims incorporated specific functional rules for computer animation that improved the existing technological process of animation. These rules enabled the automation of previously manual tasks.
Hardware-Based Improvement
A claim for a specific hardware-based RFID serial number data structure was also considered eligible. The court determined that this claim focused on improving the technological process of data encoding by providing a unique hardware-based data structure.
The USPTO provides additional clarity on AI-assisted inventions, recognizing that AI may serve as a tool for human inventors.
Key Elements of AI Patent Eligibility
- Novelty and Non-Obviousness
AI-related inventions must introduce new methods or substantial improvements beyond existing technology. This novelty ensures the AI solution is not merely a reapplication of known techniques. - Utility and Practicality
Inventions must demonstrate a clear practical use case. AI tools without defined functionality or applications are unlikely to qualify. - Beyond General AI Algorithms
The guidance emphasises that patents will only be granted for innovations offering more than standard algorithms. Inventors must focus on unique features, such as improved data processing methods or new problem-solving techniques. - Defining Patent-Eligible Subject Matter
AI inventions should highlight technical contributions, such as enhanced computational efficiency, rather than simply automating known processes. This distinction helps ensure that only impactful AI innovations receive protection.
Impact on AI Innovations
Challenges for Innovators
- Abstract Nature of AI: Many AI inventions risk being classified as abstract ideas, leading to rejection as a list of few examples may not cover the entire scope of AI related technologies.
- Strict Eligibility Criteria: It is difficult to ascertain a technical contribution by innovators that goes beyond the standard of AI functionalities.
Opportunities for Innovators
The guidance encourages innovation by defining how AI-driven processes and tools can qualify for patents. It supports AI innovators by ensuring their inventions receive protection while maintaining high standards for patent eligibility.
It helps the innovators to navigate their inventions in a better way.
Though the guidance has provided a framework for the examination of AI invention, there are still a few challenges in implementing this guidance as AI is an evergrowing technology, and therefore it is difficult to apply the same standards or follow the same guidance for every case. The challenges in implementing this guidance are:
- Complex Interpretation: Some stakeholders have expressed concerns about the lack of clarity in certain aspects of the guidance.
- Limited examples and case studies: A few examples, as given in the guidance, will not be enough for the innovators to completely understand how AI applications will be examined; therefore, more examples and case studies will be helpful for the innovators.
- Compliance and Adaptation: Smaller companies may struggle to meet the new standards due to limited legal resources.
Need for Clarity
While the 2024 Updated Guidance provides valuable insights into the patentability of AI inventions, some areas remain unclear and require further clarification:
- Specific Technical Improvements:
- The guidance emphasises the need for specific technical improvements, but it can be challenging to define and quantify what constitutes a “specific” improvement in the context of AI. Therefore, further clarification is needed on the level of detail required in the specification to define what constitutes a specific improvement.
- Balancing Abstractness and Practical Application:
- The line between an abstract idea and a practical application can be blurred in AI-related inventions.
- The guidance should provide more concrete examples to help innovators distinguish between eligible and ineligible claims.
- AI-Generated Inventions:
- As AI systems become more sophisticated, they may generate innovative ideas and solutions.
- The guidance should also address the patentability of inventions generated solely or partially by AI systems.
- Evolving Nature of AI:
- AI technology is rapidly evolving, and new techniques and applications emerge frequently.
- The guidance should be flexible enough to adapt to these advancements and avoid becoming outdated.
AI Patent Strategy
The guidance suggests two primary approaches to establishing the patentability of AI inventions. The choice of AI patent strategy often depends on the breadth of the invention’s application.
1. Focus on the Implementation Details
One strategy to establish patentability involves incorporating specific implementation details that surpass human mental capabilities while avoiding explicit references to underlying mathematical concepts.
2. Clarify the Invention’s Downstream Benefits
Another effective strategy, as demonstrated in the USPTO’s analysis of machine learning-based speech source separation, involves explicitly claiming the downstream benefits of an AI invention.
Key Takeaway
The USPTO’s new guidance on AI inventions represents a critical step toward modernising patent regulations in response to the rapid development of AI technologies. Innovators should take advantage of this guidance by focusing on technical improvements and ensuring their inventions demonstrate practical utility. While challenges remain, this guidance still offers a foundation for innovators, helping them secure valuable intellectual property rights for their AI-driven innovations.
FAQs
1. What are the key changes in the USPTO’s new guidance on patent-eligible AI inventions?
The new guidance focuses on ensuring novel, useful, and technically significant AI inventions qualify for patent eligibility while preventing the patenting of abstract AI processes.
2. How does the USPTO determine whether an AI invention is patent-eligible?
The USPTO uses a two-step framework to assess whether an AI invention is abstract and if it integrates into a practical application.
3. What challenges do innovators face under the new guidelines?
Innovators face challenges in proving that their inventions go beyond generic AI algorithms and offer unique technical contributions.
4. How can inventors best prepare their AI patent applications?
Inventors should focus on specific technical improvements and demonstrate utility to enhance their chances of success.
5. What impact does the USPTO’s AI patent guidance have on the future of innovation?
The guidance promotes innovation by clarifying patent eligibility and ensuring that only high-value AI inventions receive protection.
