- Law Firms
August 7, 2023 - In the rapidly evolving landscape of artificial intelligence (AI), innovation is at an all-time high. From machine learning algorithms that predict disease onset to neural networks that make financial predictions, AI has been instrumental in pushing technological boundaries. However, as technology accelerates, our legal systems, particularly the patent system, struggle to keep pace.
Central to patent law is the Alice/Mayo patent eligibility test. This test is designed to evaluate whether a particular invention is eligible for patent protection. The basic requirements for patent eligibility encompass "useful process, machine, manufacture, or composition of matter."
However, the U.S. Supreme Court decisions in the cases of Mayo Collaborative Services v. Prometheus Laboratories, Inc. and Alice Corp. v. CLS Bank International held that mere laws of nature, natural phenomena, and abstract ideas are not sufficient for patent eligibility. The Court emphasized that while abstract ideas or natural phenomena in isolation aren't patent eligible, their transformation into a practical application serving as an innovative building block could be.
The Alice/Mayo test, a two-part framework, aids in making this distinction. It first evaluates if claims are targeting an abstract idea, law of nature, or natural phenomenon. If they are, the test then checks for additional elements that make the claim innovative and distinct from the foundational concept.
However, when applied to AI technologies, the test can yield subjective, inconsistent and sometimes contentious results. The crux of the issue lies in the fact that many AI innovations can be viewed as abstract since they involve algorithms and mathematical processes. Determining where the line is drawn between a patent-ineligible abstract idea and a patent-eligible inventive concept in the realm of AI can be challenging.
The missed opportunities of the Supreme Court
Earlier this year, the Supreme Court had an opportunity to clarify this complex issue. Three pivotal cases — Interactive Wearables LLC v. Polar Electro Oy, No. 21-1281 (U.S. May 15, 2023); Tropp v. Travel Sentry Inc., No. 22-22 (U.S. May 15, 2023), and Avery Dennison Corp. v. ADASA Inc., No. 22-822 (U.S. May 30, 2023) — were anticipated to provide more definitive guidelines on patent eligibility.Interactive Wearables LLC v. Polar Electro Oy questioned the appropriate standard for determining whether a patent claim is directed to a patent-ineligible concept under the Alice/Mayo framework.
Tropp v. Travel Sentry Inc involved an appeal of a decision that determined that patents for a method of providing consumers with special dual-access luggage locks that a screening entity would access in accordance with a special procedure and corresponding key controlled by the luggage screening entity, all while allowing the luggage to remain locked following screening, were patent ineligible.
Avery Dennison Corp. v. ADASA Inc. questioned whether a claim for a patent to subdivide a binary serial number and assign the "most significant bits" such that they remain identical across Radio Frequency Identification Device (RFID) tags constitutes patent-eligible subject matter.
While not directly related to AI, their decisions could have potentially set precedent, or offered insights that would clarify some of the complexities encountered when dealing with AI-based inventions. These cases were seen as potential vehicles to address the ambiguities surrounding what is considered an abstract idea versus an inventive concept — a central dilemma in AI patent evaluations. By choosing not to review these cases, the Supreme Court perpetuated uncertainty, particularly for stakeholders in the AI domain.
Why AI technologies are most impacted
AI, by its very nature, often blurs the line between abstract and concrete. Abstract ideas, like mathematical formulas, are not patent-eligible. But what happens when such a formula becomes a machine-learning model driving real-world outcomes? This gray area is where AI innovations typically reside.
Specifically, while the algorithms themselves may seem abstract and theoretical, when implemented, they can drive revolutionary changes in various sectors, including health care, finance, and entertainment. For example, an AI-based system in cybersecurity within a financial network may employ a mathematical model like a neural network, which, while abstract in nature, leads to the practical application of enhanced detection of fraudulent transactions.
The inherent subjectivity in making this distinction means that AI technologies often tread a thin line between acceptance and rejection during the patent examination and evaluation process. This ambiguity can deter innovators from pursuing patents, given the unpredictability of the outcomes, thereby potentially stifling advancement.
Towards more robust AI patent claims
The onus now falls on inventors, businesses and their legal teams to craft patent claims that can stand up to this ambiguous eligibility test. Here are some key general considerations for drafting patent claims for AI-based inventions:
1. Specify the domain
Specifying the domain in AI patent claims is pivotal for multiple reasons. By clearly defining the field of application, inventors delineate the scope of their invention, ensuring protection in a specific niche. This clarity avoids overbroad claims that risk rejection and highlights the novelty within a particular context.
For patent examiners, a clear domain offers context and understanding, streamlining the examination process. Moreover, a well-defined domain strengthens both the enforcement and defense of patents while simplifying licensing and commercialization efforts. In the world of AI patents, precision in detailing the domain is important when capturing the invention's true essence and value.
2. Detail the AI mechanism
Rather than make a generic reference to "AI," highlighting the precise mechanism, be it a neural network, deep learning, or reinforcement learning, provides clarity on the technology's foundation. This granularity not only emphasizes the uniqueness of the invention but also aids patent examiners in understanding its intricacies.
A well-defined AI mechanism enhances the patent's robustness, ensuring that its distinctiveness is clearly captured and protected. In the realm of AI patents, specificity in describing the AI mechanism is a linchpin for safeguarding innovation.
3. Highlight technical advantages
Emphasizing the technical advantages of the AI in patent claims is essential. When a claim articulates clear benefits, such as improved efficiency or error reduction, the intrinsic value of AI's contribution to the invention becomes evident. This not only underscores the invention's uniqueness but also showcases its practical significance. In the patenting of AI technologies, focusing on these technical advantages can bolster the patent's strength, clearly differentiating it from mere abstract ideas or generic applications.
4. Avoid over-reliance on algorithms
Though algorithms form the backbone of AI, an excessive focus on them might overshadow their real-world applications. It's essential to strike a balance: outline the algorithm's essence but pivot to its tangible applications and results. This approach not only captures the innovation's full spectrum but also strengthens its patentability by emphasizing its practical impact over mere theoretical constructs.
Tips for drafting AI-based patent claims
Given the nuanced nature of AI and the intricacies of patent law, specific strategies can optimize the chances of obtaining robust patent protection. Here are some essential tips for drafting AI-based patent claims, each addressing distinct facets of the patenting process to ensure that AI innovators secure the recognition and protection they deserve.
1. Multi-layered claiming
Using a multi-layered claiming approach in AI patent applications is strategic. Beginning with a broader claim and then transitioning to more specific, dependent claims ensures a comprehensive coverage. This layered strategy acts as a safety net; if a broad claim faces rejection, the subsequent, more detailed claims might still secure approval. In essence, it's a way to cast a wide net while also having targeted catches, optimizing the chances of patent protection across varying depths of the invention's scope.
2. Functional claiming
Opting for functional claiming in AI patent applications emphasizes the tool's practical utility over its internal workings. By focusing on what the AI tool accomplishes, like "identifying anomalies in X data," instead of delving into the intricate algorithmic steps, the claim is anchored in tangible outcomes.
This not only simplifies the claim's language but also broadens its protective scope, covering potential variations in algorithmic implementations that achieve the same functional result. In the AI patent landscape, functional claiming offers a way to capture the essence of an invention's real-world impact.
3.Include data specificity
In AI patent applications, data specificity is integral. Given AI's intrinsic reliance on data, delineating the data type, its processing method and its significance to the invention offers a clearer picture of the technology in action. By anchoring the claim in the specifics of the data utilized, the invention's distinctiveness and practical utility are underscored.
In essence, detailing the data landscape not only clarifies the AI tool's operation but also strengthens the patent's grounding in tangible and innovative use cases.
4. Avoid over-generalizing
Steering clear of over-generalization in AI patent claims is essential. While casting a wide net might seem attractive, exceedingly broad claims risk being tagged as abstract and face rejection. It's more prudent to focus on the distinct facets and practical applications of the invention. Such focused claims not only stand a better chance of securing patent protection but also effectively highlight the value proposition and innovation the AI tool brings to the table.
5. Stay updated
Staying abreast of developments is crucial in the fluid landscape of AI patenting. Given the swift advancements in software and AI, coupled with shifting patent regulations, it's vital to keep tabs on current case law, patent office directives and industry shifts. Being informed ensures that your patent strategies remain aligned with the latest legal precedents and can adapt to emerging trends, maximizing the chances of securing robust and relevant protection for AI innovations.
In conclusion, drafting patent claims for AI-based inventions requires a careful blend of technical detailing, legal foresight and strategic breadth. By ensuring patent eligibility and crafting a precise claim language, inventors can significantly enhance the chances of their AI tool being protected by a patent.
Opinions expressed are those of the author. They do not reflect the views of Reuters News, which, under the Trust Principles, is committed to integrity, independence, and freedom from bias. Westlaw Today is owned by Thomson Reuters and operates independently of Reuters News.
Anup Iyer
Anup Iyer is an associate with Moore & Van Allen. He specializes in assisting clients with obtaining patent and trademark rights across diverse technology sectors such as artificial intelligence (AI), optical communication, high performance computing, computer processor architecture, wireless communication technologies, and cybersecurity. He is based in Charlotte, North Carolina, and may be reached at anupiyer@mvalaw.com.
Nick Russell
Nick Russell is a member with the firm, based out of the Charlotte office. He provides patent portfolio management and guidance to institutions in the network science sectors, financial sectors, and health care sectors. He has extensive experience in patent preparation and prosecution in computer and network science technologies, payment networks, authentication methods, cybersecurity, cloud computing, and mobile device technology. He may be reached at nickrussell@mvalaw.com.
FAQs
Drafting patent claims for AI-based inventions: navigating eligibility and precision? ›
In conclusion, drafting patent claims for AI-based inventions requires a careful blend of technical detailing, legal foresight and strategic breadth. By ensuring patent eligibility and crafting a precise claim language, inventors can significantly enhance the chances of their AI tool being protected by a patent.
How to draft patent claims for machine learning inventions? ›- Focus on the structure of the ML model in the claim.
- Identify if the invention lies in the training phase or execution phase or both.
- Claim the training process.
- Emphasize the input data preparation.
- Cover the input mapping to the model.
Current Patent Law on AI Inventorship
Under current patent law, only a "person" may be an inventor, and an inventor must be a natural person. AI cannot be named as an inventor because it is not a legal person. The current law also requires that an inventor contribute to the conception of the invention.
The USPTO and Federal Circuit's opinions imply that inventions made by human beings with the assistance of AI are eligible for patent protection. However, there is uncertainty and much debate as to how much AI assistance is too much for patentability.
Can AI write a patent application? ›By leveraging advanced machine learning algorithms, generative AI can create bespoke writing content comparable in quality to that produced by human drafters, presenting a significant advancement in the automation of patent drafting.
How do you write an algorithm patent? ›According to U.S. patent law, you cannot directly patent an algorithm. However, you can patent the series of steps in your algorithm. That's because an algorithm is seen as a series of mathematical steps and procedures under U.S. patent law. Note: Most people confuse machine software and software patents.
How do you write an invention claim? ›Writing a basic patent claim
Every claim has three sections—the preamble, the transitional phrase, and the body of the claim. The preamble is the first part of the claim. In the writing instrument claim above, the preamble is the phrase “A writing instrument for making a mark on a writing surface”.
As of December 2022, Baidu was the largest owner of active machine learning and artificial intelligence (AI) patent families worldwide with 13,993 active patent families owned. In 2022, the company had claimed the leading position from Tencent now ranked second with 13,187 active patent families owned.
Who may be the owner of an invention created by AI? ›In the case of AI-generated inventions, the human(s) that used AI to devise such inventions should be named as the inventor in most cases. The identification of the inventor is significant because the right to own a patent, and therefore its benefits, flows from the named inventor.
What is patentable in AI? ›Broadly speaking, an AI invention is only excluded from patent protection if it does not reveal a technical character or serve a technical purpose. An AI invention is likely to make a technical contribution if, when it runs on a computer, its instructions: Embody a technical process which exists outside the computer.
Can I sell AI generated art as my own? ›
Can you sell AI-generated art? Yes, AI-generated art can be sold just like any other artwork. In fact, there is a growing market for AI art, with some pieces selling for significant amounts of money. Here is a summary of the most popular styles of AI art which can be sold online.
Should AI have patent rights? ›Allowing such AI inventions to be patented would encourage “the owners of these systems to disclose inventions rather than keep them as trade secrets, and it will encourage the investment needed to commercialize inventions,” Abbott said.
Can an AI claim copyright? ›At the moment, works created solely by artificial intelligence — even if produced from a text prompt written by a human — are not protected by copyright.
How many patent applications are there for AI? ›How Many AI Patents Are There? The number of AI patents has surged over the last few years; there are roughly 18,753 as of 2021. The largest increase in AI patents filed was in 2022, recording the highest average annual growth rate (AAGR) of 28%.
How many AI related patent applications are there? ›The number of patents issued by the US Patent & Trademark Office for AI technologies has surged over the past five years. During this timeframe, the number of AI-related patents issued has increased from 3,267 in 2017 to 18,753 in 2021.
Can I copyright my AI generated art? ›In other words, anything that comes out of an AI program can't be protected under copyright law and will not be accepted even if it's included in a work created by a human.
Can you patent an ML algorithm? ›A machine learning algorithm must satisfy the requirements for patentability, which include invention, non-obviousness, and usefulness, in order to be qualified for a patent. This implies that the method must be novel, not readily apparent to an expert in the topic, and have some value or practical application.
Can you patent a learning method? ›Like all other patents, a patent for an educational process must satisfy three basic elements: the invention must be 1) useful, 2) non-obvious, and 3) novel (new). If these requirements are met, then an educational process is likely eligible for patent protection.