Navigating AI Patent Applications to Avoid Section 101 Rejections

It feels like AI is becoming more popular and ubiquitous with each passing day.  This naturally means that more and more people and companies are looking to patent AI-based inventions that cover more and more types of technologies.  However, as the intersection of artificial intelligence (AI) and patent law grows increasingly complex, these people and companies may find themselves in a patent-eligibility minefield. Particularly, there is a risk of an AI-based patent application being deemed an abstract idea and rejected under 35 U.S.C. § 101. Below, we will discuss the following points:

1. Framework for Thinking about Section 101 Rejections
2. What Leads to Rejection?
3. What Can Help Overcome Rejection?
4. Critical Steps in the USPTO’s Analysis

What Are Section 101 Rejections: Framework for Thinking about Section 101 Rejections

Although this post is not intended to cover all the complexities of Section 101 (of which there are many!), we provide a quick overview for context as it relates to AI-specific issues.  At its core, the question of Section 101 is whether an invention amounts to an abstract idea that is merely implemented by a computer.  In the Alice case, which is a 2014 Supreme Court case from which the modern law on this issue is derived, the patents were related to a computer-implemented scheme for mitigating “settlement risk” (i.e., the risk that only one party to a financial transaction will pay what it owes) by using a third-party intermediary (i.e., escrow).  In that case, the Supreme Court found that the patents were patent-ineligible because their central concept is the abstract idea of intermediated settlement, and merely implementing that abstract idea on a computer is insufficient to take it out of the realm of “abstract.”

In many instances, AI-based patent applications relate to software, and therefore, the question of patent eligibility under Alice is relevant.  In other words, AI-based patent applications are often a subset of software-based patent applications.

The Alice case provided a test for determining whether a patent claim is patent-eligible under Section 101.  The chart below provides a (simplified) reference guide.  The intricacies of this chart are outside the scope of this post, but the chart does provide a helpful visual reference for thinking through Section 101 issues as they relate to AI-based patent applications.

Avoiding the Pitfalls: What Leads to Rejection?

When drafting a patent application for an invention that involves AI, it is best to avoid elements that may be construed as abstract ideas without practical application.

The USPTO often identifies claims that involve purely mathematical concepts or mental processes as abstract ideas. These are seen as non-patentable because they do not add any substantive technical improvement or innovation.

For instance, an application that merely recites a process of using a neural network to perform calculations or detect patterns without any specific technological application or improvement is likely to be rejected as abstract. The USPTO may view these types of claims as too broad and merely instructing one to apply the abstract idea using a generic computer, which fails to meet the patent eligibility requirements under Section 101.

Enhancing Your Chances: What Can Help Overcome Rejection?

To increase the likelihood of your patent application being accepted, it’s helpful to demonstrate that your invention applies AI technology in a concrete and innovative way. Claims should integrate the AI-driven abstract idea into a practical application or show that the AI system improves the functionality of a particular technology.

For instance, demonstrating how an AI innovation enhances network security by detecting and preventing malicious attacks in real time can establish its practical application. Clearly outlining how AI drives technological improvements, like boosting security or efficiency, strengthens claims against Section 101 rejections.

Critical Steps in the USPTO’s Analysis: Where Rejections Are Made or Avoided

Step 2A Prong One: The USPTO examines whether your claims recite an abstract idea. This hurdle will often be easy for examiners to satisfy. It is often difficult to reframe an invention to avoid such characterizations.

For example, an examiner might argue that, under the broadest reasonable interpretation, your claim covers the steps that could practically be performed in the human mind but using generic computer components.

This includes using a fundamental economic concept or managing interactions between people.

Simply describing mathematical relationships, formulas, or calculations without practically, concretely, and/or tangibly being applied to solving an actual problem in the real world is not sufficient. In short, you cannot patent E=MC2, but you can patent a nuclear bomb.

Step 2A Prong Two: The USPTO then considers whether the claim integrates the abstract idea into a practical application. Demonstrating that your invention applies AI in a specific, meaningful way to solve a technical problem can help your application pass this prong.

This is your best chance to turn a Section 101 rejection into an allowance. If your claim is too broad, the examiner will repeat what they said in their Step 2A Prong One rejection. While an examiner may concede that your claim may include “additional elements,” these elements are often dismissed by repeating the examiner’s previous characterization of your claim as mere instructions to use generic computer components or the like.

Step 2B: Finally, if the claim is deemed to involve an abstract idea, the USPTO assesses whether there is an “inventive concept” that transforms the abstract idea into a patent-eligible invention. This step can be pivotal if your application shows that the AI method contributes significantly more than just the abstract idea, thereby justifying patentability.

If your claim has not survived any of the previous steps, it is unlikely to be saved in Step 2B. Usually, the same analysis as Step 2B will be applied. If you can successfully overcome Step 2B, you can overcome Step 2A Prong Two.  This may be preferable because it is often easier to clearly describe the technical details of your invention rather than getting drawn into a vague, philosophical discussion of what constitutes an “inventive concept.”

Understanding these steps and crafting your patent claims to address them directly can make the difference between a successful and rejected patent application.

Tips for a Successful Patent Application

Examiners may reject your AI invention for various reasons, some of which can be avoided with well-drafted claims. Others require refining the claim language or presenting compelling arguments. Avoid encountering these common USPTO objections.

Claims should be written to prevent the patent examiner from characterizing your invention in the following ways.

BAD:

• You claimed the “idea” of a solution rather than an actual solution.
• You “simply applied” AI to automate something.
• Your claim is equivalent to a generic instruction to “apply” an abstract idea.
• Your claim performs a mathematical calculation that could be practically performed in the human mind.
• Your claim is just data gathering activities. Examiners call this a “well-known extra-solution activity.”
• Your claim does not “meaningfully limit” the claim beyond a mere indication of the field of use or technological environment in which the abstract idea is performed.

Instead, your claims should support the patent examiner characterizing your invention in one of the following ways.

Pro Tip: Envision someone reviewing your application cold, without additional context. How easily could they challenge your claim? Consider what concise argument they might use to justify allowing it to their supervisor.

GOOD:

• Your claimed AI model and training data provide a “particular way to achieve a desired outcome.” Be specific about both the method and the outcome.
• Your claim includes specific technical details about the AI model. Most AI inventors leverage existing models, such as Gemini, Llama, and countless others, and apply them to custom datasets or integrate them into tailored workflows or interfaces.
• Your claim should include precise details about the AI model training process. Most small AI inventors use pre-trained models without direct control over or insight into the training, avoiding the need for extensive resources.
• Your claim includes technical details about how the AI model is integrated into a particular method to improve conventional approaches.
• Your claim includes a neural network model trained using a plurality of training datasets.
• Your claim includes a training dataset with at least two data types associated with each other. For example, a dataset of patient records or medical information might include physiological parameters related to a patient’s disease indicator.
• Your claim “meaningfully limits” the application of the abstract idea. This guidance is vague but important. The USPTO gives the example of claiming to select something specific from a list of specifically enumerated items rather than a broad category.

Conclusion

Understanding and addressing Section 101 rejections is paramount in navigating the complex landscape of AI patent applications. Inventors can effectively distinguish their innovations from abstract ideas by crafting claims emphasizing specific technological applications and meaningful improvements. A clear and strategic approach mitigates rejection risks and strengthens the foundation for securing intellectual property rights in the evolving intersection of AI and patent law. With a proactive mindset and attention to detail, applicants can ensure their inventions contribute to technological advancement while meeting the rigorous standards of the USPTO. To learn more about how the experienced attorneys at NK Patent Law can assist you in navigating the complexities of AI inventions and patent applications, contact us today.

About NK Patent Law

NK Patent Law is an intellectual property boutique serving clients throughout the United States and abroad.  We focus on patent preparation and prosecution, trademark preparation and prosecution, and general IP counseling to help businesses navigate complex IP issues and potential disputes. We help startups, established companies, and universities protect their intellectual property and navigate the competitive landscape to protect their valuable intellectual property assets. Visit us at www.nkpatentlaw.com to learn more.