The U.S. Patent and Trademark Office recently issued new guidance on the eligibility of AI-related inventions for patenting, focusing on determining when claims involving artificial intelligence (AI) can be considered suitable for patent protection. Released on July 17, 2024, the "2024 Patent Subject Matter Eligibility Guidance Update Including on Artificial Intelligence" (2024 AI SME Update) provides crucial insights for patent prosecutors navigating the complexities of AI patent applications.

This guidance introduces three sets of claim examples to demarcate the boundary between eligible and ineligible claims concerning AI inventions. It aims to offer clarity, particularly in cases where patent eligibility hinges on the interpretation of hardware components and software processes in AI-related technologies.

The Office's guidance pertains to the frequently encountered Section 101 rejections, dealing with subject matter eligibility under 35 U.S.C. 101. These rejections involve a rigorous two-step analysis to ascertain whether a claim addresses patentable subject matter. Initially, it is necessary to evaluate if the invention fits into any of the patentable categories: processes, machines, manufactures, or compositions of matter. Subsequently, the Supreme Court’s Alice/Mayo framework is applied to ascertain whether the claim is a judicial exception, such as laws of nature, natural phenomena, or abstract ideas, and if additional elements within the claim are deemed inventive.

The newly released examples (Examples 47-49) highlight these evaluative processes. In particular, Example 47 provides two contrasting claims highlighting varied outcomes in patent eligibility:

Claim 1: Eligible Patent Claim

Claim 1, which involved an application-specific integrated circuit (ASIC) for an artificial neural network (ANN), was deemed eligible for patent protection. This outcome was primarily due to the claim's focus on specific hardware components, effectively circumventing the abstract idea limitation by grounding the invention in a tangible, physical form. The detailed description of the ASIC’s hardware architecture, including its neurons and synaptic circuits, was pivotal in establishing its eligibility.

Claim 2: Ineligible Patent Claim

Conversely, Claim 2, which focused on a method of using an ANN, was found ineligible. It was primarily seen as resting on abstract ideas and general data processing, deemed insubstantial without detailed innovation in its implementation on a computing platform. The use of conventional AI training methodologies and generic computer functions did not persuasively demonstrate a technological improvement or inventive concept that could elevate it beyond a mere mental process or abstract mathematical operation.

Claim 3: Eligible Patent Claim

Claim 3, akin to Claim 2 in content but differing in application, was eligible due to its practical application in improving network security. The claim tied AI-driven methodologies to a defined technological improvement by demonstrating enhanced real-time network intrusion detection capabilities, a crucial factor under the subject matter eligibility framework.

The guidance underscores the vitality of drafting claims that integrate AI innovations within practical applications to not only meet patent eligibility standards but also display technical advancements over existing technologies. For patent applicants involved in AI innovation, these examples provide critical benchmarks to guide claim drafting towards patentability. As AI continues to permeate various technological fields, navigating these guidelines becomes essential in securing intellectual property rights within the legal frameworks governing patents.

Source: Noah Wire Services