Navigating Patent Challenges for AI Software and Hardware

Artificial Intelligence (AI) is transforming industries at an unprecedented rate, leading to a surge in patent filings for both AI-driven software applications and AI-specific hardware, such as AI chips. However, patenting AI innovations comes with significant legal and technical challenges that differ depending on whether the invention lies in software or hardware. Understanding these differences is crucial for innovators seeking robust intellectual property (IP) protection.

Patenting AI Software: The Challenges of Abstract Ideas

AI software primarily involves algorithms, machine learning models, and data processing techniques that power applications in industries like healthcare, finance, and autonomous systems. However, patenting AI software can be challenging due to subject matter eligibility concerns under 35 U.S.C. §101 in the United States. Courts and the U.S. Patent and Trademark Office (USPTO) often reject AI-related software claims as “abstract ideas” unless they demonstrate a significant technical improvement.

Key Challenges for AI Software Patents:

1. Subject Matter Eligibility: AI software must be framed as a practical application that improves computing processes, rather than merely reciting an algorithm.

2. Obviousness Issues (35 U.S.C. §103): Many AI techniques, such as neural networks and deep learning, are considered well-known, making it difficult to prove their novelty and non-obviousness.

3. Enablement & Written Description (35 U.S.C. §112): AI models are often complex and may require extensive disclosures to satisfy enablement requirements, especially if they rely on large datasets for training.

To enhance patent eligibility, AI software innovations should be claimed in a way that ties the invention to specific improvements in computing technology. For example, instead of claiming a general AI model, a patent application may focus on a unique method for optimizing training efficiency or reducing computational overhead in neural networks.

Patenting AI Hardware: Navigating Semiconductor and Processing Innovations

AI hardware encompasses specialized processors, such as AI chips, designed to accelerate AI computations. Unlike software, AI hardware inventions are generally considered patent-eligible because they involve tangible, physical components. AI chips, such as Graphics Processing Units (GPUs), Tensor Processing Units (TPUs), and neuromorphic processors, present unique opportunities for patent protection.

Key Considerations for AI Hardware Patents:

1. Patent Eligibility Advantage: AI hardware typically avoids the “abstract idea” hurdle since it involves a physical device, making it easier to secure patent protection.

2. Overcoming Obviousness: Many AI hardware innovations involve incremental improvements in semiconductor design, which can face §103 rejections. Demonstrating significant performance gains or novel architectures can help overcome these challenges.

3. Thermal Management and Power Efficiency: Many AI chips focus on energy efficiency and heat dissipation. Innovations in these areas may be strong candidates for patent protection due to their critical role in high-performance AI applications.

4. Interoperability and Standard-Essential Patents (SEPs): AI chips that integrate with existing computing architectures may be subject to standard-essential patent (SEP) considerations, requiring strategic licensing approaches.

Comparing AI Software and AI Hardware Patents

For AI companies, a balanced patent strategy is essential:

• For AI software, focus on demonstrating specific technical improvements in computing operations to navigate §101 issues.

• For AI hardware, ensure that patent applications highlight unique circuit designs, power efficiency improvements, or innovative semiconductor structures to establish novelty.

• Hybrid AI innovations, where software and hardware function together (e.g., AI accelerators with specialized software frameworks), may benefit from a combination of hardware and method claims to maximize protection.

As AI continues to evolve, securing strong patent protection for both software and hardware will be key to maintaining a competitive edge in the market. By understanding the distinct challenges and opportunities in each domain, innovators can build a robust AI patent portfolio that safeguards their technological advancements.

USPTO issues AI subject matter eligibility guidance

On July 16, 2024, the United States Patent and Trademark Office (USPTO) introduced revised guidelines for assessing patent applications, particularly those involving critical and emerging technologies such as artificial intelligence (AI). This guidance aims to clarify the application of the Alice/Mayo test, a two-part framework used to determine whether a patent claim is directed to eligible subject matter under 35 U.S.C. § 101. The test evaluates if a claim involves a “judicial exception” such as an abstract idea, a law of nature, or a natural phenomenon, which are generally not eligible for patenting.

The USPTO's two-part Alice/Mayo test begins by determining if the claim is directed to a judicial exception. If it is, the analysis proceeds to whether the claim integrates the exception into a practical application. If the exception is integrated, the claim is eligible. If not, the USPTO then assesses if the claim includes additional elements that amount to significantly more than the exception. Claims that do so are eligible; otherwise, they are rejected. The newly released guidance aims to provide clearer examples and updated analysis to help examiners and applicants navigate this test, particularly for AI-related inventions.

The updated guidelines emphasize that AI inventions are typically seen as computer-implemented inventions and offer more clarity on how these inventions can be patented. For instance, AI claims that improve the functioning of a computer or another technology are more likely to be considered eligible. The guidance also highlights the importance of clearly describing technological improvements in the patent specification. Claims that specify a particular way to achieve a desired outcome, rather than broadly stating the outcome, are more likely to be deemed eligible.

Additionally, the USPTO has introduced three new Subject Matter Eligibility Examples providing additional analyses under 35 U.S.C. § 101 of hypothetical claims in specific situations. These examples address particular inquiries such as whether a claim recites an abstract idea or integrates the abstract idea into a practical application. The examples include using neural networks to detect data anomalies, AI methods for analyzing speech signals, and AI models for personalizing medical treatments. They are intended to assist USPTO personnel in applying the USPTO’s subject matter eligibility guidance to AI inventions during patent examination, appeal, and post-grant proceedings. These examples serve as a practical guide for drafting and prosecuting patent applications, showing that claims are more likely to be eligible if they demonstrate a specific technological improvement.

The guidance, effective July 17, 2024, will accept public comments through September 16, 2024. Whether future court decisions will align with this guidance remains uncertain. However, this development marks a significant step in clarifying patent eligibility for emerging technologies, ensuring that inventors receive adequate protection for their innovations. Ortiz & Lopez, PLLC will continue to monitor these changes to provide our clients with the most current and effective patent prosecution strategies.