The U.S. Patent and Trademark Office (USPTO) released Updated Guidance concerning the patentability of AI-assisted inventions that provides detailed analysis of how to assess subject matter eligibility of AI-assisted inventions under 35 U.S.C. 101 of the Patent Act.
While the legal test for analyzing subject matter eligibility, explained in detail in the Manual of Patent Examining Procedure Section 2106, has not changed, the updated guidance is intended to provide yet further assistance in the rapidly growing field of AI technology.
The updated guidance explains that "whether an invention was created with the assistance of AI is not a consideration" in determining whether it is patent-eligible. The inquiry instead is focused on the claimed subject matter itself. As explained in the Prior-Issued USPTO Guidance, discussed in our article USPTO Issues Inventorship Guidelines for AI-Assisted Inventions, patentability is not barred for AI-assisted inventions that were created with "significant contribution" from one or more human co-inventors.
The updated guidance explains that "a key point of distinction to be made for AI inventions" lies with the claims themselves. Specifically, when the claims are directed to an improvement in technology, they constitute patent-eligible subject matter. On the other hand, when the claims are "no more than instructions to apply an abstract idea" using a general computer or computer system components, or only generally link an idea to a particular technological environment, the claims are not patent eligible.
The updated guidance refers to McRO, Inc. v. Bandai Namco Games America Inc., 837 F.3d 1299 (Fed. Cir. 2016), which concerned patents for AI-assisted technology that provides lip synchronization of computer-animated characters. In that decision, the Federal Circuit held the claimed invention patent eligible because it utilized AI to solve or at least improve the technological problem of the characters' lip synchronization.
Explanatory Examples
The updated guidance provides three examples of hypothetical AI-implemented inventions that explain whether exemplary claims for such inventions are patent eligible under the current legal test.
- Detection of Network Intrusions Using AI Neural Networks
In this example, claims are presented that recite an application-specific integrated circuit (ASIC) implementation of an artificial neural network for anomaly detection. According to the analysis, the claims that recite a specific organization of the neuron arrays and hardware (i.e., not software) are readily considered as not being directed to an abstract idea and thus are deemed patent eligible. The broader method claim, which is directed to a process for detecting network intrusions, while deemed to recite an abstract idea in the analysis, akin to a mental process for identifying network intrusions, also is deemed patent eligible in the analysis because the claimed process integrates the abstract idea into a practical application that improves network security technology.
- AI for Analyzing Speech Input Signals
This example involves an AI speech model to process audio input and separate the desired speech from background noise or extraneous speech. The exemplary broad claim that recites simply identifying (using AI) the speech portions in the audio input is assessed to be an abstract idea, without more, and thus is deemed not patent eligible. On the other hand, the exemplary claims that recite incorporating the speech analysis AI model into an improved speech signal processing method and computer system are assessed to be patent eligible. In this example, a key takeaway is that the AI model is not simply being applied merely in general to the field of speech recognition. Instead, the AI model is utilized in a "specific way" by generating a specific mask to improve speech recognition processing in a noisy environment. Accordingly, as illustrated in the example, specificity and practical integration help make claims patent eligible even if the underlying AI model itself is deemed to represent an abstract idea.
- AI for Personalized Medical Treatment
The final example is an invention that uses AI to personalize medical treatment in connection with individual characteristics of a specific patient. The broader exemplary claim is assessed to be a non-patentable mental process because it calls for using an AI model to recommend treatment for a patient based only on patient information and treatment outcomes. As explained, the broad claim itself does not integrate the abstract idea into a practical application or provide any limitation that is deemed significantly more than the abstract idea. In contrast, the other exemplary claim, which recites using a more specific AI model based on a neural network and training data to select a treatment from a specific list of options, is assessed to be patent eligible since it particularly integrates the abstract idea into a practical application.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.