The Federal Circuit's recent decision in Recentive Analytics, Inc. v. Fox Corp. (April 18, 2025) has garnered a lot of attention. This is not surprising: It hits on hot topics such as machine learning, artificial intelligence (AI) and patent eligibility, and the court itself called this a "matter of first impression."
But the decision should not be surprising to those who follow patent eligibility jurisprudence.
The Core Issue: Generic Machine Learning Applications vs. Technological Improvements
The Federal Circuit affirmed a lower court's dismissal of the patent infringement claims, concluding that patents merely applying generic machine learning techniques to new data environments – without meaningful technological improvements to the machine learning models themselves – are ineligible under 35 U.S.C. § 101.
The plaintiff asserted two different sets of patents. Looking at the machine learning training patents, the representative claim recited a method that claimed: "(i) a collecting step (receiving event parameters and target features); (ii) an iterative training step for the machine learning model (identifying relationships within the data); (iii) an output step (generating an optimized schedule); and (iv) an updating step (detecting changes to the data inputs and iteratively generating new, further optimized schedules."
Though focusing on machine learning, these patents are not unlike the mass of patents invalidated since Alice because the patent claims were directed to collecting, analyzing, updating and/or displaying information.
The fact that the analysis step included machine learning was not going to save these patents, especially when the claim language was generic. The Federal Circuit, looking at both the claims and the specification, found that the "technology described in the patents is conventional." The Federal Circuit highlighted a fundamental problem:
"Recentive's own representations about the nature of machine learning vitiate [its] argument: Iterative training using selected training material and dynamic adjustments based on real-time changes are incident to the very nature of machine learning."
In other words, the patents claimed using a machine learning model in the generic sense. That is not enough.
The Missing "How": A Common Flaw
In determining whether a patent claim is directed to an abstract idea or a technological improvement, courts often consider whether the claim includes how to achieve the purported improvement. The Federal Circuit did the same here:
Neither the claims nor the specifications describe how such an improvement was accomplished. That is, the claims do not delineate steps through which the machine learning technology achieves an improvement.
…
Allowing a claim that functionally describes a mere concept without disclosing how to implement that concept risks defeating the very purpose of the patent system.
This "functional claim language, without more" was deemed "insufficient for patentability" under established precedent. And this analysis – the question of "how" – is common, so it is not surprising that these claims failed here.
But, though common, that does not mean it is without controversy. The same issue caused intense disagreement at the Federal Circuit a handful of years ago in the American Axle v. Neapco decision. As in this case, Judge Timothy Dyk wrote that the asserted claims did not include the necessary how. Judge Kimberly Moore, in dissent, raised concerns with eligibility under Section 101 being conflated with enablement under Section 112. (Read more in our previous blog, "A Federal Circuit Quarrel: Patent Eligibility, Enablement and a Fiery Dissent.")
Why This Matters for AI and Machine Learning Patent Strategy
This ruling underscores an important distinction for those seeking to patent AI/machine learning applications:
- Simply applying existing machine learning techniques to new domains (such as television scheduling in these patents) is likely insufficient.
- Patent applications must articulate specific technological improvements to the underlying machine learning methods.
The court explicitly stated that "machine learning is a burgeoning and increasingly important field and may lead to patent-eligible improvements in technology." It distinguished, however, between genuine technical advances and mere applications of existing technology.
For companies developing machine learning solutions:
- Focus patent strategies on technical improvements to machine learning algorithms themselves, not just their application to new fields.
- Clearly articulate how claimed innovations improve upon conventional machine learning approaches.
- Avoid purely functional claiming that describes what the AI does without explaining the technical means of achievement.
- Consider complementary protection strategies (trade secrets, copyright) for applications that may face Section 101 challenges.
AI innovation is accelerating. This decision provides another reminder that merely applying existing techniques to new domains will not pass patent eligibility muster.
