In an era where artificial intelligence is reshaping every industry, the line between patent-eligible inventions and unpatentable abstract ideas has become increasingly blurred. In this article, I review the USPTO’s July 2024 guidance update on AI patent eligibility and see if it helps to clarify the murky waters of AI intellectual property.

Introduction

In July 2024, the United States Patent and Trademark Office (USPTO) released an important update to its guidance on patent subject matter eligibility, with a particular focus on artificial intelligence (AI) inventions. This update, effective July 17, 2024, aims to clarify the application of 35 U.S.C. § 101 to AI-related patent claims in light of the rapidly evolving technological landscape. The new guidance builds upon the existing Alice/Mayo two-step framework and introduces three new examples specifically addressing AI inventions. For founders and executives of early-stage tech companies, understanding these updates is crucial for developing effective IP strategies in the AI era.

July 2024 Updated Eligibility Guidance

Key Points of the USPTO’s AI Eligibility Guidance

1. The guidance, effective July 17, 2024, aims to clarify subject matter eligibility for AI inventions under 35 U.S.C. § 101. It emphasizes that AI-assisted inventions are not categorically unpatentable.
2. The guidance applies the existing Alice/Mayo two-step framework to AI inventions. It addresses two main concerns for AI inventions:
   • Evaluating whether a claim recites an abstract idea (Step 2A, Prong One)
   • Assessing improvements to technology (Step 2A, Prong Two)
3. The USPTO introduced three new examples specifically for AI-related inventions:
   • Example 47: AI neural network for anomaly detection
   • Example 48: AI-based speech signal analysis
   • Example 49: AI model for personalized medical treatment

These examples illustrate both patent-eligible and ineligible claims to guide applicants and examiners.

Takeaways

1. AI use in invention doesn’t automatically make it unpatentable or ineligible.
2. Focus on demonstrating how your AI invention improves computer functionality or solves a technological problem.
3. Clearly articulate the practical application of your AI technology in patent claims.

The Alice/Mayo Two-Step Framework

The Alice/Mayo test, established by the Supreme Court decisions in Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012) and Alice Corp. v. CLS Bank International (2014), provides a two-step framework for determining patent eligibility under 35 U.S.C. § 101:

Step 1: Determine whether the claim is directed to a patent-ineligible concept.

This step asks whether the claim is directed to a law of nature, natural phenomenon, or abstract idea. If not, the claim is eligible. If yes, proceed to Step 2.

Step 2: Determine whether the claim recites additional elements that amount to “significantly more” than the judicial exception.

This step looks for an “inventive concept” – elements or a combination of elements that ensure the patent in practice amounts to significantly more than a patent upon the ineligible concept itself.

Important considerations in applying this framework include:

1. Abstract ideas can include mathematical concepts, certain methods of organizing human activity, and mental processes.
2. “Significantly more” can be demonstrated by:
   • Improvements to the functioning of a computer or to any other technology or technical field
   • Applying the judicial exception with, or by use of, a particular machine
   • Effecting a transformation or reduction of a particular article to a different state or thing
   • Adding a specific limitation other than what is well-understood, routine and conventional in the field
3. Merely implementing an abstract idea on a generic computer or adding insignificant extra-solution activity is not enough to confer eligibility.
4. The analysis must consider the claim as a whole, rather than dissecting it into individual elements.

This framework is designed to strike a balance between protecting true innovation and preventing the monopolization of fundamental scientific and technological principles. However, its application, particularly to software inventions, has been the subject of ongoing debate and refinement. There has been a lot written about patent eligibility of software inventions, and I will not address the pros and cons of the Alice/Mayo framework here. Instead, let’s review the new AI examples offered in the USPTO’s updated guidance.

Examples

Here is a summary of the three new AI-related examples provided in the guidance update. The full text of the examples can be found here.

Example 47: Anomaly Detection

This example illustrates how AI-specific claims, particularly those involving artificial neural networks (ANNs) for anomaly detection, are evaluated for subject matter eligibility.

Key Points:

1. Claim 1 (eligible): Describes a specific hardware implementation of an ANN (an ASIC with specific neuron and synaptic circuit configurations). This claim is eligible as it falls within a statutory category and doesn’t recite any judicial exceptions.

[Claim 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.

2. Claim 2 (ineligible): Recites a method of using an ANN for anomaly detection, including steps of data preprocessing, training, and analysis. This claim is ineligible because it recites abstract ideas (mathematical concepts and mental processes) without integrating them into a practical application or providing significantly more.

[Claim 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.

3. Claim 3 (eligible): Builds on Claim 2 but adds specific steps for detecting and responding to malicious network packets. This claim is eligible because it integrates the abstract idea into a practical application by improving network security.

[Claim 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.

Implications:

• Hardware implementations of AI systems are more likely to be eligible.
• Merely reciting AI processes without practical applications may be insufficient for eligibility.
• Integrating AI-based analysis into specific technological improvements can lead to eligibility.

Example 48: Speech Separation

This example focuses on AI methods for separating speech signals from mixed audio inputs.

Key Points:

1. Claim 1 (ineligible): Describes a method of processing mixed speech signals using a deep neural network (DNN) to determine embedding vectors. It’s ineligible because it recites mathematical concepts without integration into a practical application.

[Claim 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 temporal features extracted from the mixed speech signal x; and

(c) using a deep neural network (DNN) to determine embedding vectors V using the formula V = fθ(X), where fθ(X) is a global function of the mixed speech signal x.

2. Claim 2 (eligible): Builds on Claim 1 by adding steps to synthesize and combine speech waveforms, excluding a target source. This claim is eligible because it integrates the abstract idea into a practical application of speech separation technology.

[Claim 2] The speech separation method of claim 1 further comprising:

(d) partitioning the embedding vectors V into clusters corresponding to the different sources sn;

(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 sn;

(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 ss such that the mixed speech signal x’ includes speech waveforms from the different sources sn and excludes the speech waveform from the target source ss; and

(h) transmitting the mixed speech signal x’ for storage to a remote location.

3. Claim 3 (eligible): Describes a computer-readable medium with instructions for speech separation and transcription. It’s eligible because it integrates the abstract idea into a practical application of speech-to-text conversion.

[Claim 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 sn, 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 source ssn; 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 sd.

Implications:

• Mere data processing using AI, even with complex algorithms, may not be sufficient for eligibility.
• Practical applications of AI in audio processing, especially those that solve technological problems, can lead to eligibility.
• Combining AI processing with specific output generation (like transcription) can strengthen eligibility arguments.

Example 49: Fibrosis Treatment

This example illustrates the application of AI in personalized medical treatments.

Key Points:

1. Claim 1 (ineligible): Describes a method of identifying patients at high risk of post-implantation inflammation using an AI model and administering an appropriate treatment. It’s ineligible because it recites abstract ideas and laws of nature without integrating them into a practical application.

[Claim 1] A post-surgical fibrosis treatment method comprising:

(a) collecting and genotyping a sample from a glaucoma patient to a provide a genotype dataset;

(b) identifying the glaucoma patient as at high risk of post-implantation inflammation (PI) based on a weighted polygenic risk score that is generated from informative single-nucleotide polymorphisms (SNPs) in the genotype dataset by an ezAI model that uses multiplication to weight corresponding alleles in the dataset by their effect sizes and addition to sum the weighted values to provide the score; and

(c) administering an appropriate treatment to the glaucoma patient at high risk of PI after microstent implant surgery.

2. Claim 2 (eligible): Specifies the treatment as Compound X eye drops. This claim is eligible because it integrates the abstract idea into a practical application by applying a particular treatment.

[Claim 2] The method of claim 1, wherein the appropriate treatment is Compound X eye drops.

Implications:

• AI-based diagnostic methods alone may not be eligible without specific treatment steps.
• Combining AI-based analysis with particular treatments can lead to eligibility.
• The specificity of the treatment matters in determining eligibility.

Overall, these examples demonstrate that while AI-related inventions often involve abstract ideas or mathematical concepts, they can be eligible when they:

1. Implement AI in specific hardware configurations.
2. Integrate AI analysis into practical applications that solve technological problems.
3. Combine AI-based diagnostics with specific treatment methods.

The USPTO is clearly trying to provide guidance on how AI inventions can be patent-eligible, emphasizing the importance of practical applications and technological improvements beyond mere data processing or analysis.

Conclusion

The USPTO’s 2024 guidance update is helpful in providing clearer guidelines and concrete examples of how to craft patents directed to inventions involving the use of AI. The key to success lies in understanding the nuances of the guidance and strategically framing AI inventions to emphasize their practical applications and technological improvements. As AI continues to advance at a breakneck pace, those who can effectively navigate these patent eligibility guidelines will be best positioned to protect their innovations and maintain a competitive edge in the market.

Moving forward, it’s crucial for tech companies to work closely with patent attorneys who are well-versed in these new guidelines. By doing so, they can ensure that their AI innovations are not only groundbreaking but also patent-eligible.