On April 18, 2025, the Court of Appeals for the Federal Circuit (CAFC) ruled in Recentive Analytics Inc. v. Fox Corp. et al. that new uses for established machine learning do not make the claims patent-eligible.
Recentive Alleges Fox’s Infringement of Machine Learning Patents
Recentive owns four patents involving the use of machine learning to arrange and schedule broadcasts. For example, the NFL uses Recentive’s software to schedule broadcasts. In this case, Recentive sued Fox alleging that Fox uses software for regional scheduling that infringes Recentive’s patents.
How Patent Eligibility Analysis Applies to Software
All patent applications undergo a patent eligibility analysis under Title 35 USC 101. Patentable subject matter has four categories:
- Process
- Machine
- Apparatus
- Composition of matter
For most patent applications, this makes for a simple analysis. Some subject matter exists that may not be afforded patent protection, including abstract ideas.
Software applications take a different journey through this analysis. Software encompasses a process, so the first part of the analysis finds software patentable. However, software may also be found to be an abstract idea, which triggers a further analysis under the US Supreme Court’s 2014 ruling in Alice Corp. v. CLS Bank International.
One way to overcome the finding of an abstract idea involves showing that the abstract idea comprises “something more” than the abstract idea itself.
The CAFC said Recentive’s patents cover only the abstract idea of using machine learning to generate event schedules and TV broadcasts, finding the patents to cover patent-ineligible subject matter. This does not mean that all machine learning patents cover ineligible subject matter. Here, the CAFC could have easily reached the same decision under a different part of the patent law.
Recentive’s Position: Known Machine Learning, New Application in Broadcasting
In this case, Recentive acknowledged that the concept of creating “network maps” for arranging TV broadcasts has existed for a long time. Prior to computers, humans used to handle scheduling. Recentive also conceded that the machine learning method was not novel and that their patents solely covered applying machine learning to TV scheduling. Considering these facts, it hardly comes as a surprise that the CAFC ruled that there wasn’t more to the patent’s abstract idea than the abstract idea itself, one of the definitions of patent-ineligible subject matter. Indeed, if the subject matter were not machine learning, it could just as easily have been found to be obvious—regardless of whether it covered machine learning or not.
In addition to meeting the requirements for patent-eligible subject matter, patent claims also need to cover subject matter that is new and nonobvious. “New” subject matter means that no one has patented it before. “Nonobvious” subject matter means that the invention covered by the claims would not be obvious to someone skilled in the art, even if there was no single patent that covered the same subject matter.
New Applications of Machine Learning Don’t Equate to New Innovation
Machine learning is a tool, like an axe or a hammer. Using the same axe to chop down an oak tree after the same axe has been used to chop down a pine tree does not make the axe itself patentable. It is still an axe. Applying known technology to a new application has always given rise to questions of obviousness. In some cases, the new use involves enough unique elements to render the patent valid. However, using a known technology for a different application does not render the subject matter of the patent nonobvious. This holds especially true cases like Recentive’s, where the integration of the known technology did not give rise to anything new or novel. Moreover, the claims of these patents did not even describe any specifics about the machine learning technology beyond simply stating that it was machine learning.
In summary, the ruling found the machine learning patents to cover ineligible subject matter. They could just have easily been found to be obvious. The facts do not indicate that this means all machine learning patents will suffer the same fate, just those that use the machine learning axe to cut down a different tree.
This article is provided for informational purposes only—it does not constitute legal advice and does not create an attorney-client relationship between the firm and the reader. Readers should consult legal counsel before taking action relating to the subject matter of this article.
