By Erik Swanson
In Recentive Analytics v. Fox Corp. (April 18, 2025) the Federal Circuit affirmed the District of Delaware’s dismissal on patent-ineligibility grounds of a suit seeking to enforce patents directed to scheduling and network‑map optimization in the broadcasting context. The Federal Circuit held the patents, some of which asserted uses of machine learning (ML) to generate live-event schedules and others to generate network programming maps, to be patent ineligible. Recentive marks the first appellate decision directly confronting whether applying generic ML techniques to a novel field can satisfy §101 of the U.S. patent statute.
Recentive’s patents
Recentive’s patents were directed to solving problems confronting the entertainment industry and television broadcasters: how to optimize the scheduling of live events and how to optimize “network maps,” which determine the programs or context displayed by a broadcaster’s channels at given time within specific geographic markets. The scheduling patents claimed computer-implemented methods comprising, among other steps, “generating, via [a] trained ML model, a schedule for the future series of live events that is optimized relative the one or more prioritized event target features” based on a detected “real time change” in one or more user-specific parameters so that the schedule dynamically remains optimized. The network map patents claimed computer-implemented methods comprising, among other steps, “using a machine learning technique to optimize an overall television rating across [a] first plurality of live events and [a] second plurality of live events” and “automatically updating the network map on demand and in real time based a change to at least one of (i) [a] schedule [of the live events] and (ii) underlying criteria.”
The Federal Circuit’s patent eligibility analysis
To assess whether Recentive’s patents pass muster under §101, the Federal Circuit applied the analytical framework set forth in the U.S. Supreme Court’s 2014 Alice Corp. v. CLS Bank decision. Under the Alice two-step analysis, the first step is to determine whether the claims at issue are directed to one of the three judicial exceptions to patent-eligibility (laws of nature, physical phenomenon, or abstract ideas). If the claims are directed to a judicial exception, the second step is to determine whether additional elements of the claim are sufficient to transform the nature of the claim into an eligible application of the judicial exception. Additional elements can suffice to transform the claim where there is an “inventive concept” or otherwise ensures that the claim in practice amounts to “significantly more” than a patent on an ineligible concept.
Under Alice Step 1, the court deemed Recentive’s claimed inventions to be “directed to the abstract idea of using a generic machine learning technique in a particular environment,” noting Recentive’s own admissions that the patents neither claimed new ML algorithms nor improvements to existing techniques. The court pointed to statements in the patents themselves specifying that “any suitable machine learning technology … such as, for example, a gradient boosted random forest, a regression, a neural network, a decision tree, a support vector change, a Bayesian network,” maybe be used to carry out the invention. The patents fail to specify any non-generic computing architecture or any novel improvement to algorithmic performance or training techniques, the court noted. The court concluded that “the only thing the claims disclose about the use of machine learning is that machine learning is used in a new environment.” The asserted innovations—iteratively training models to optimize schedules or maps—were consequently held to be nothing more than the abstract idea of conventional data-processing methods being applied to broadcasting, lacking any specific improvement to computing capabilities.
Proceeding to Step 2, the court found no “inventive concept” that would transform the abstract idea into a patent-eligible application. The use of iterative training and dynamic updating was inherent to ML itself and offered no unconventional implementation. As such, the application of ML to optimize schedules and maps was described as more of the abstract idea rather than an inventive concept. The patents were further criticized for reciting generic computing hardware, offering no unique integration of the ML steps into a technological improvement. The court concluded that the patent claims amount to “no more than claiming the abstract idea itself.”
Conclusion/key takeaway
The Recentive decision solidifies the existing trend in Federal Circuit §101 jurisprudence: limiting abstract ideas to a particular filed or implementing them in generic computing devices is insufficient to confer patent eligibility. As such, it serves as a reminder to patent applicants and practitioners that, particularly in the realm of AI and ML, patents need to demonstrate a tangible technological improvement or inventive concept beyond the mere application of existing techniques. It is thus not enough to merely deploy ML; patent-eligibility depends on a demonstrated technical advancement.
Recentive has filed a petition for rehearing by the court en banc, arguing that its decision does away with the line between § 101 and §§ 102/103, and improperly imports § 112 enablement issues into the § 101 inquiry.
