On April 18, 2025, the U.S. Court of Appeals for the Federal Circuit issued a precedential opinion in Recentive Analytics, Inc. v. Fox Corp.1 The Federal Circuit held that the Asserted Patents2 — which relate to methods of generating real-time network maps and optimized event schedules using machine learning — were directed to patent-ineligible subject matter under 35 U.S.C. § 101. The claims were held to be patent-ineligible because they were directed to the abstract idea of using a generic machine learning technique in a particular environment, with no inventive concept.
Recentive Analytics is the first Federal Circuit case to meaningfully address patent-eligibility under Section 101 for AI-based patents (in this case machine learning). In its analysis, the Federal Circuit relied heavily on Recentive’s concessions from its briefing, oral argument, and the specifications of the Asserted Patents to conclude that the claimed machine learning technology was generic and conventional. As a consequence, the Federal Circuit’s opinion purports to resolve only a narrow question on patent-eligibility of AI-based patents: “[t]oday, we hold only that patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under [Section] 101.”3 However, the underlying analysis indicates that this opinion may extend beyond the narrow question the Federal Circuit claims to have resolved, which may have broad-ranging implications on the patentability analysis of AI-based patents going forward.
In this article, we provide a summary of the Federal Circuit’s opinion and its consequences. We conclude this article with our suggestions for practitioners to best protect AI-based inventions in the wake of the Recentive Analytics decision.
Case Summary
In 2022, Recentive sued Fox for patent infringement over the Asserted Patents in the U.S. District Court for the District of Delaware.4 Fox moved to dismiss for failure to state a claim on the ground that the Asserted Patents were directed to patent-ineligible subject matter under Section 101. The district court granted the motion, and Recentive appealed to the Federal Circuit.
The Asserted Patents were directed toward optimizing the scheduling of live events and how to optimize “network maps,” to determine programming broadcasted within certain geographic markets at certain times.
At the first step of the Alice inquiry,5 the Federal Circuit held that claims of the Asserted Patents were clearly directed to abstract ideas — namely, generating event schedules and network maps.
In its analysis, the Federal Circuit pointed to concessions from Recentive that the Asserted Patents did not claim machine learning itself and only employed generic/conventional machine learning technology. Further, the Federal Circuit took issue with the claims of the Asserted Patents because they “do not delineate steps through which the machine learning technology achieves an improvement.”6
Recentive tried to establish that the claimed machine learning methods were not generic or conventional, arguing the claimed methods employed iterative training and dynamic adjustment based on real-time changes to input data, which were technological improvements to conventional machine learning models. The Federal Circuit did not find these arguments persuasive because the “[i]terative training using selected triaining material and dynamic adjustments based on real-time changes are incident to the very nature of machine learning.”7 The Federal Circuit was also unpersuaded by Recentive’s arguments that the claimed machine learning models were patent-eligible because they are implemented in a particular technological environment and/or improve the speed and efficiency of tasks previously undertaken by humans, citing direct conflict with the Federal Circuit’s established Section 101 jurisprudence.8,9
At the second step of the Alice inquiry, the Federal Circuit found that the claims lacked an inventive concept and did not transform the abstract idea into a patent-eligible application. Recentive argued that the inventive concept of the claims lied in the dynamic output feature of the machine learning model based on real-time data. However, the Federal Circuit determined this feature was nothing more than claiming the abstract idea itself. The Federal Circuit concluded that because the Asserted Patents failed both steps of the Alice inquiry, the claims were not directed to patent-eligible subject matter and affirmed the district court’s decision.
Takeaways and Trends
Recentive Analytics provides at least three takeaways practitioners should consider when litigating or prosecuting patents involving AI and machine learning.
Patent Claims Should Focus on Algorithm Method Over End-Use Benefit
First, if the patent is not conferring an improvement to computing components or the operation of devices themselves, the Federal Circuit is adopting the view that AI-based claims must recite specific method steps that describe how a machine learning model’s algorithm is configured to be patent-eligible. In other words, for an AI-based patent method using machine learning to be patent-eligible, the claims should focus on the algorithm itself, rather than mere recitation of the downstream benefit achieved by using AI. Further, without specificity around the improvements to the machine learning model or functionality of the components, utilizing machine learning in a new environment is not likely sufficient to render claims patent-eligible.
Real-Time Data and Dynamic Adjustments Alone May Not Be Technological Improvements
Second, using real-time data and dynamic adjustments to the machine learning model may not be enough to render claims patent-eligible. Here, Recentive argued that those exact features were the technological improvements to conventional machine learning models. Seemingly, the Federal Circuit’s holding indicates that simply applying machine learning models using real-time data would not be a technical improvement to a machine learning model, without reciting specific method steps describing how the machine learning algorithm achieves this.
Federal Circuit and PTAB Alignment May Indicate Tougher Path for AI Patents
Third, the Federal Circuit’s analysis and outcome largely mirrors that which we have seen at the Patent Trial and Appeal Board (“PTAB”) following the United States Patent and Trademark Office’s updated guidance on patent eligibility analysis of AI-based inventions issued in July 2024.10 For instance, in October 2024, the PTAB determined that the claims in Ex parte Ma were patent-ineligible under Section 101.11 The claims in Ex parte Ma related to computer-implemented methods of using a series of neural networks and a discriminative model that processed earning call transcripts to output a prediction on stock price movement. The claims were held to be patent-ineligible because they were directed to an abstract idea, the claimed AI models were considered generic and conventional, and the claims lacked an inventive concept because any conceivable technological improvements were nothing more than improvements in efficiency for tasks previously undertaken by humans. Parity between the PTAB and the Federal Circuit in their Section 101 analysis of AI-based inventions/patents may be indicative of an uphill battle for inventors of AI-based inventions/patents.
Conclusion
Only time can provide clarity on the extent to which the decision in Recentive Analytics will impact AI-based patents. While the dust settles, patent practitioners may want to consider taking the following precautions when working with AI-based inventions and patents:
- For applications that implement an AI method as a component of the claim, consider drafting applications to avoid inclusion of language that could have the AI method construed as generic or conventional. For example, avoid declaring that an AI model is not particularly limited or specifying the name of AI models that can be implemented within a claimed method (e.g., random forest, regression, neural network, etc.). Instead, try to describe AI models functionally, in as much detail as possible, while emphasizing unique algorithmic steps that differentiate or distinguish the inventive AI model from existing convention models.
- To the extent possible, try to draft applications that tie the AI method to an improvement of the computer (or model) itself, rather than how it improves a downstream, otherwise human task. For example, if appropriate, consider adding disclosure that the AI method requires less training data, less training time, increased accuracy, or other features that demonstrate advancements over the prior art.
- Thoroughly discuss the underlying details of the AI methods your clients are using. In instances where those models are conventional or known in the art, consider advising clients toward other avenues of intellectual property protection, such as copyright or trade secret.
[1] Recentive Analytics, Inc. v. Fox Corp., Fox Broadcasting Company, LLC, Fox Sports Productions, LLC, Case No. 23-2437, (Fed. Cir. 2025).
[2] U.S. Patent Nos. 10,911,811 (“’811 Patent”), 10,958,957 (“’957 Patent”), 11,386,367 (“’367 Patent”), and 11,537,960 (“’960 Patent”).
[3] Recentive Analytics, Inc. v. Fox Corp., Fox Broadcasting Company, LLC, Fox Sports Productions, LLC, Case No. 23-2437, (Fed. Cir. 2025) at 18.
[4] Recentive Analytics, Inc. v. Fox Corp., 692 F. Supp. 3d 438 (D. Del. 2023).
[5] Alice Corporation v. CLS Bank International, 573 U.S. 208 (2014).
[6] Recentive Analytics, Inc. v. Fox Corp., Fox Broadcasting Company, LLC, Fox Sports Productions, LLC, Case No. 23-2437, (Fed. Cir. 2025) at 13.
[7] Id. at 12.
[8] Citing Intell. Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363 (Fed. Cir. 2015) and SAP Am., Inc. v. InvestPic, LLC, 898 F.3d 1161 (Fed. Cir. 2018) for Section 101 jurisprudence on claims implementing abstract ideas in novel, particular technological environments being held patent-ineligible.
[9] Citing Content Extraction & Transmission LLC v. Wells Fargo Bank, Nat’l Ass’n, 776 F.3d 1343 (Fed. Cir. 2014), Trinity Info Media, LLC v. Covalent, Inc., 72 F.4th 1355 (Fed. Cir. 2023), and Customedia Techs., LLC v. Dish Network Corp., 951 F.3d 1359 (Fed. Cir. 2020) for Section 101 jurisprudence on claims improving speed and efficiency compared to humans being held patent-ineligible.
[10] 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence, 89 Fed. Reg. 58128 (July 17, 2024) (AI Guidance).
[11] Appeal No. 2023-003593, U.S. App. No. 17/030,953.
