USPTO July 2024 Patent Eligibility Guidance for AI Inventions

Introduction:

On July 16, 2024, the United States Patent and Trademark Office (USPTO) released updated guidance on patent subject matter eligibility, specifically addressing artificial intelligence (AI) inventions. This guidance was issued in accordance with Executive Order 14110, which emphasizes the safe, secure, and trustworthy development and use of AI. The primary objective of the guidance is to assist USPTO personnel and stakeholders in evaluating the patent eligibility of AI-related claims and to provide clarity and consistency in the examination process. It is effective July 17, 2024.

Background:

The necessity for updated guidance stems from the increasing role of AI in technological innovation and the associated challenges in determining patent eligibility. The USPTO has actively engaged with stakeholders through various initiatives and public consultations to understand the unique considerations AI inventions pose for patent eligibility. Previous efforts, including the 2019 Revised Patent Subject Matter Eligibility Guidance (2019 PEG) and subsequent updates, have laid the groundwork for this comprehensive guidance update.

Key Components of the Guidance

1. Evaluating AI Inventions:
   The guidance emphasizes the application of the USPTO’s subject matter eligibility framework to AI inventions, ensuring that these inventions are evaluated consistently with other types of computer-implemented inventions. It clarifies how to apply the two-step framework established by the Supreme Court in Alice Corp. v. CLS Bank International and Mayo Collaborative Services v. Prometheus Laboratories, Inc.
2. Step 1 – Determining Statutory Subject Matter:
   In the first step, examiners must determine whether the claimed invention falls into one of the four statutory categories: processes, machines, manufactures, or compositions of matter. This step sets the boundaries of the claim’s coverage and influences whether the claim encompasses subject matter beyond these categories.
3. Step 2 – The Alice/Mayo Test:

• Step 2A (Prong One): This step involves determining whether the claim recites a judicial exception, such as an abstract idea, law of nature, or natural phenomenon. For AI inventions, this often involves assessing whether the claim includes mathematical concepts, methods of organizing human activity, or mental processes.
• Step 2A (Prong Two): If a claim recites a judicial exception, the next step is to determine whether the claim integrates the exception into a practical application. This involves evaluating whether the additional elements in the claim transform the judicial exception into a patent-eligible application by imposing a meaningful limit on the exception.
• Step 2B: If the claim is directed to a judicial exception, the final step is to evaluate whether the additional elements in the claim provide an inventive concept that amounts to significantly more than the judicial exception itself. This includes considering whether the additional elements are well-understood, routine, or conventional activities in the field.

Practical Application and Improvements

The guidance highlights the importance of demonstrating that an AI invention improves the functioning of a computer or another technology. This can be achieved by showing that the claimed invention addresses a specific technological problem with a technological solution, thereby integrating the judicial exception into a practical application. The USPTO provides examples and case law to illustrate how this evaluation should be conducted.

Examples and Case Law

The guidance includes new examples (Examples 47-49) specifically designed to assist examiners in applying the subject matter eligibility framework to AI inventions. These examples cover various scenarios and technologies, providing practical insights into how claims should be evaluated for patent eligibility. The guidance also references relevant Federal Circuit decisions to support the analysis of AI-related claims.

Detailed Analysis of New Examples

Example 47: Anomaly Detection

Claim 1: Eligible

1. An application-specific integrated circuit (ASIC) for an artificial neural network (ANN), the ASIC comprising:

• a plurality of neurons organized in an array, wherein each neuron comprises a register, a microprocessor, and at least one input;
• and a plurality of synaptic circuits, each synaptic circuit including a memory for storing a synaptic weight, wherein each neuron is connected to at least one other neuron via one of the plurality of synaptic circuits.

Reasoning: Falls within statutory category as a machine. No judicial exceptions are recited. The claim focuses on a specific hardware (circuit) designed to facilitate neural network operations, distinguishing it from general-purpose computing hardware.

Claim 2: Ineligible

2. A method of using an artificial neural network (ANN) comprising:

• (a) receiving, at a computer, continuous training data;
• (b) discretizing, by the computer, the continuous training data to generate input data;
• (c) training, by the computer, the ANN based on the input data and a selected training algorithm to generate a trained ANN, wherein the selected training algorithm includes a backpropagation algorithm and a gradient descent algorithm;
• (d) detecting one or more anomalies in a data set using the trained ANN;
• (e) analyzing the one or more detected anomalies using the trained ANN to generate anomaly data; and
• (f) outputting the anomaly data from the trained ANN.

Reasoning: Although it falls under the statutory category as a process, it recites judicial exceptions (abstract ideas) related to mathematical concepts used in data processing and training an ANN. The process fails to integrate these exceptions into a practical application and does not add an inventive concept beyond the abstract idea itself.

Claim 3: Eligible

3. A method of using an artificial neural network (ANN) to detect malicious network packets comprising:

• (a) training, by a computer, the ANN based on input data and a selected training algorithm to generate a trained ANN, wherein the selected training algorithm includes a backpropagation algorithm and a gradient descent algorithm;
• (b) detecting one or more anomalies in network traffic using the trained ANN;
• (c) determining at least one detected anomaly is associated with one or more malicious network packets;
• (d) detecting a source address associated with the one or more malicious network packets in real time;
• (e) dropping the one or more malicious network packets in real time; and
• (f) blocking future traffic from the source address.

Reasoning: This claim incorporates a judicial exception but is considered integrated into a practical application. It improves network security through specific, non-generic steps that utilize an ANN to perform real-time identification and remediation actions based on the detection of anomalies, which is a technical improvement over prior methods.

Example 48: Speech Separation

Claim 1: Ineligible

1. A speech separation method comprising:

• (a) receiving a mixed speech signal x comprising speech from multiple different sources sn, where n ∈ {1, … N};
• (b) converting the mixed speech signal x into a spectrogram in a time-frequency domain using a short time Fourier transform and obtaining feature representation X, wherein X corresponds to the spectrogram of the mixed speech signal x; and
• (c) using a deep neural network (DNN) to determine embedding vectors V using the formula V = f_θ(X), where f_θ(X) is a global function of the mixed speech signal x.

Reasoning: The method recites a judicial exception by using mathematical formulas for signal processing. It lacks integration into a practical application, merely applying an abstract idea using a generic digital neural network without a specific technological improvement or inventive concept.

Claim 2: Eligible

2. The speech separation method of claim 1 further comprising:

• (d) partitioning the embedding vectors V into clusters corresponding to the different sources s_n;
• (e) applying binary masks to the clusters to create masked clusters;
• (f) synthesizing speech waveforms from the masked clusters, wherein each speech waveform corresponds to a different source s_n;
• (g) combining the speech waveforms to generate a mixed speech signal x‘ by stitching together the speech waveforms corresponding to the different sources sn, excluding the speech waveform from a target source s_s such that the mixed speech signal x’ includes speech waveforms from the different sources s_n and excludes the speech waveform from the target source s_s; and
• (h) transmitting the mixed speech signal x’ for storage to a remote location.

Reasoning: Extends beyond Claim 1 by effectively integrating the judicial exception into a practical application. It specifies how speech signals are processed, separated, and synthesized using defined algorithms, contributing to the technological field by enhancing the clarity and separation of speech components from complex audio signals.

Claim 3: Eligible

3. A non-transitory computer-readable storage medium having computer-executable instructions stored thereon, which when executed by one or more processors, cause the one or more processors to perform operations comprising:

• (a) receiving a mixed speech signal x comprising speech from multiple different sources s_n, where n ∈ {1, … N}, at a deep neural network (DNN) trained on source separation;
• (b) using the DNN to convert a time-frequency representation of the mixed speech signal x into embeddings in a feature space as a function of the mixed speech signal x;
• (c) clustering the embeddings using a k-means clustering algorithm;
• (d) applying binary masks to the clusters to obtain masked clusters;
• (e) converting the masked clusters into a time domain to obtain N separated speech signals corresponding to the different sources s_n; and
• (f) extracting spectral features from a target source sd of the N separated speech signals and generating a sequence of words from the spectral features to produce a transcript of the speech signal corresponding to the target source s_d.

Reasoning: Though it involves mathematical calculations, the claim integrates these into a practical application by detailing a complete process for improving speech-to-text accuracy using specialized clustering and masking techniques, which represent technical advancements in audio processing.

Conclusion

The eligibility of AI-related claims largely hinges on whether the use of AI and related mathematical processes can be shown to contribute a specific, technical improvement or a novel and non-obvious application. Claims that merely recite these technologies without such integration tend to be ineligible under current guidelines, highlighting the importance of clearly demonstrating how AI contributes beyond conventional computer functions or abstract ideas. Each of these examples illustrates different facets of this principle, with the eligible claims showing clear, technical benefits and specific applications of the underlying AI technologies.