On April 18, 2025, the U.S. Court of Appeals for the Federal Circuit (the main court for resolving patent issues) issued a significant decision in Recentive Analytics, Inc. v. Fox Corp. (No. 23-2437), affirming the district court’s dismissal of Recentive’s patent infringement claims against Fox. The court held that Recentive’s patents, which utilized generic machine learning techniques for event scheduling and network mapping, were not patentable, since they were directed to abstract ideas and lacked the inventive concept necessary for patent eligibility under 35 U.S.C. § 101.
The Patents in Question
Recentive owned four patents—U.S. Patent Nos. 10,911,811; 10,958,957; 11,386,367; and 11,537,960—grouped into two categories:?
- Machine Learning Training Patents: Focused on optimizing live event schedules using machine learning models trained on historical data.?
- Network Map Patents: Aimed at creating optimized broadcasting schedules by generating network maps based on various data inputs.?
The claims involved applying generic machine learning to steps like collecting event parameters, training machine learning models, and generating optimized schedules or network maps.?
The Court’s Analysis
Applying the two-step framework from Alice Corp. v. CLS Bank International, the Federal Circuit concluded:
- Abstract Idea: The patents were directed to abstract ideas—specifically, the use of generic machine learning techniques for scheduling and mapping tasks.?
The court emphasized that merely applying generic machine learning techniques to a particular field does not make the claims patentable.? The Court also held that simply using machine learning to speed up activity traditionally done by humans is not patentable.
- Lack of Inventive Concept: The claims did not introduce any inventive concept that transformed the abstract ideas into patent-eligible applications.?
The Court also found that the claims did not include “an element or combination of elements”
that transformed the claims into something “significantly more” than applying generic machine learning to a generic problem.
Implications for Patent Filings
This decision underscores the importance of claim drafting in patents covering new applications of artificial intelligence. Gone are the days when judges were unaware that AI is simply a tool that one may apply to many generic problems. Today, AI patents must clearly identify a specific technological improvement or inventive concept. As this case shows, even if you do get a patent on a generic AI application to a generic problem, it is unlikely you will ever be able to enforce it. Drafting an AI patent that passes Patent Office muster and is defensible in court takes not only planning, skill, and expertise in the field of AI-patentability, but also a solid understanding of the constantly-changing AI-patentability legal landscape.
Recent Comments