NVG IP News

Intellectual Property Current Events

Machine Learning Patents Lose Eligibility in Lawsuit Backfire

May 5, 2025 | Northern Virginia Graphics, Inc.

On April 18, 2025, the Federal Circuit Court of Appeals affirmed the dismissal of a lawsuit for patent infringement filed by Recentive Analytics, Inc. against Fox Corporation. In determining the dismissal, the court concluded that Recentive’s patents were ineligible.

Recentive alleged infringement of four patents directed to applying machine‑learning techniques to event scheduling and broadcast network‑map creation. These patents fell into two groups: “Machine Learning Training” (U.S. Pat. Nos. 11,386,367 and 11,537,960), which claim methods for generating live‑event schedules via a generic machine‑learning model; and the “Network Map” (U.S. Pat. Nos. 10,911,811 and 10,958,957), which claim methods for generating and updating broadcast network maps using machine learning.

Although Recentive emphasized that its inventions dynamically customize outputs and surface “useful patterns” buried in data, the court observed that the patents admit they do not improve the underlying machine learning algorithms—rather, they merely apply the machine learning technique to the context of event scheduling and network‑map creation.

The court applied the two‑step Alice framework, concluding that the asserted claims are directed to the abstract idea of the production of network maps and event schedules, using “known generic mathematical techniques.” When looking for an inventive concept beyond simply “using machine learning to dynamically generate optimized maps and schedules based on real‑time data and update them based on changing conditions,” the court found none.

The Alice/Mayo two‑step test

Under 35 U.S.C. § 101, an invention must fall into one of the four statutory categories (process, machine, manufacture, composition of matter) and must not claim a judicial exception—laws of nature, natural phenomena, or abstract ideas—unless the claim integrates that exception into a practical application amounting to “significantly more.” The Supreme Court’s framework (originating from Alice Corporation v. CLS Bank International) breaks this inquiry into two steps:

Step 1: Determine whether the claim is directed to a judicial exception. Is the claim directed to a law of nature, a natural phenomenon, or an abstract idea?

Step 2: If the result from Step 1 is yes, evaluate whether the claim’s additional elements, individually or as an ordered combination, add “significantly more” to the exception. This “inventive concept” may lie in novel steps that integrate the exception into a practical application, improvements to computer technology, or unconventional components that ensure the claim amounts to significantly more than the judicial exception itself.

This two‑step test is the sole test for § 101 eligibility; other considerations now serve only as clues within this framework.¹

Takeaway for practitioners: merely applying off‑the‑shelf machine‑learning techniques to a new field of use—without disclosing how to improve those techniques—remains ineligible under § 101. To clear Alice’s hurdle, a patent must claim more than generic ML usage; it must reveal a specific technical innovation in the model, its training process, or its integration with computing hardware.