USPTO Issues Further Guidance on AI-Related Patent Eligibility

The update provides clarity on how to apply the Alice/Mayo test to AI innovations and includes examples of how the test applies in concrete cases.

On July 17^th, the USPTO issued a guidance update to help USPTO personnel and those who interact with the agency evaluate the subject matter eligibility of claims in patent applications involving artificial intelligence (AI). It walks through the well-documented Alice/Mayo test and adds certain details about using it to evaluate AI-related claims.

The update (“2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence”) was released in accordance with President Biden’s Executive Order 14110, which outlined the federal government’s suggested approaches and programs for the development and use of AI.

The USPTO also issued a set of “Subject Matter Eligibility Examples,” which gives concrete examples of claims and how each should be assessed in light of the update. It considers three examples:

• Anomaly Detection: Includes claims directed at the use of neural networks to detect malicious network packets, and specifically, the use of AI-based methods to enhance the functionality of the models over time.
• Speech Separation: Includes claims directed to identifying and separating extraneous or background speech from speech signals for further analysis and processing, and how AI can be used to refine operations or generate novel solutions.
• Fibrosis Treatment: Includes methods for using AI models to create personalized treatments for fibrosis patients, and how AI can provide new ways for users to interact with systems or applications.

The document includes a chart, reproduced below, that summarizes how each example meets or does not meet criteria in the Alice/Mayo test.

In the anomaly detection example, the USPTO highlights considerations such as whether the invention improves the algorithm’s performance (e.g., enable it to get better at identifying malware over time or use less processing resources to achieve the same result) and how the use of the algorithm achieves a specific, practical benefit (such as enabling a network to become safer or more secure, or recommending remediation actions).

In the speech separation example, considerations that impact patentability include how AI is applied to optimize industrial or business processes and the extent to which those optimizations solve problems more effectively than existing methods.

In the fibrosis treatment example, the guide describes how AI technologies can be applied to enhance user interactions (such as AI-driven chatbots or adaptive interfaces) that personalize user experience based on behavior and preferences. In these cases, the application should focus on the innovative aspects of the user interface that are enhanced or enabled by AI, and the novel ways in which the interface adapts to user inputs or preferences. Specifics include the use of natural language processing, pattern recognition, or adaptive learning systems.