Reference
Federal Circuit decisions on artificial intelligence patent eligibility
Federal Circuit opinions since Alice that decided whether a claim to machine learning or artificial intelligence is eligible for a patent under 35 U.S.C. § 101, with what was claimed, the result, and the court’s reason in one sentence. Newest first.
| Decided | Case | What was claimed | Result | Why |
|---|---|---|---|---|
| July 7, 2026 | Dental Monitoring SAS v. Align Technology, Inc. No. 2024-2270 (Fed. Cir. July 7, 2026); U.S. Pat. Nos. 11,049,248 and 10,755,409Nonprecedential · Lourie, J. · from N.D. Cal. (Alsup, J.) · Developments note | A deep learning neural network, trained on images, that analyzes photographs of a patient's teeth to judge how an orthodontic aligner fits and to guide image capture. | Ineligible | The claims collect, analyze and display information and apply generic machine learning to that idea in a new setting; the patents say the network can be any well-known one, and training it on particular images does not make it less generic. |
| Mar. 11, 2026 | Trustees of Columbia University in the City of New York v. Gen Digital Inc. Nos. 2024-1243, 2024-1244 (Fed. Cir. Mar. 11, 2026); U.S. Pat. Nos. 8,601,322 and 8,074,115Precedential · Dyk, J. · from E.D. Va. (Lauck, C.J.) · Developments note | Detecting malicious program behavior by running the program in an emulator and comparing its function calls to an anomaly-detection model, including a combined model built from models created on different computers. | Remanded | The claims are directed to comparing data against a model to decide whether it is anomalous, and spreading the work across computers is itself abstract. The court held the claims abstract at step one and sent back only whether the claimed model feature was conventional. |
| Feb. 24, 2026 | Rensselaer Polytechnic Institute v. Amazon.com, Inc. Nos. 2024-1725, 2024-1739 (Fed. Cir. Feb. 24, 2026); U.S. Pat. No. 7,177,798Nonprecedential · Dyk, J. · from N.D.N.Y. (Sannes, J.) · Developments note | Answering a user's natural-language request by applying case-based reasoning, which the court described as, in the context of this patent, a type of machine learning or artificial intelligence, to a database of metadata. | Ineligible | Following Recentive, a conventional application of an established AI technique to a new field, using generic database components, is abstract, and the novelty of the field supplies no inventive concept. |
| Apr. 18, 2025 | Recentive Analytics, Inc. v. Fox Corp. 134 F.4th 1205 (Fed. Cir. 2025), No. 2023-2437; U.S. Pat. Nos. 10,911,811, 10,958,957, 11,386,367 and 11,537,960Precedential · Dyk, J. · from D. Del. (Williams, J.) · Developments note | Generic machine learning models, trained on historical event data, that produce schedules for live events and maps assigning broadcasts to television stations, updating as conditions change. | Ineligible | Claims that do no more than apply generic machine learning to a new data environment, without claiming an improvement to the machine learning models themselves, are directed to an abstract idea and add no inventive concept. |
| Mar. 25, 2021 | In re Board of Trustees of the Leland Stanford Junior University 991 F.3d 1245 (Fed. Cir. 2021), No. 2020-1288; U.S. Application No. 13/486,982Precedential · Reyna, J. · from the Patent Trial and Appeal Board · Developments note | A computerized statistical method that builds a hidden Markov model to infer which parent each genetic variant came from across unrelated individuals. | Ineligible | The claims are directed to mathematical calculation and statistical modeling, and a gain in the accuracy of the calculation improves the math, not a technological process. |
| Mar. 11, 2021 | In re Board of Trustees of the Leland Stanford Junior University 989 F.3d 1367 (Fed. Cir. 2021), No. 2020-1012; U.S. Application No. 13/445,925Precedential · Reyna, J. · from the Patent Trial and Appeal Board · Developments note | A computerized method that uses a hidden Markov model to determine which parent each genetic variant was inherited from. | Ineligible | The claims are drawn to abstract mathematical calculations and statistical modeling; storing and providing the result is not a practical application, and better math on a generic computer is not a technological improvement. |
| Dec. 28, 2015 | Vehicle Intelligence & Safety LLC v. Mercedes-Benz USA, LLC No. 2015-1411 (Fed. Cir. Dec. 28, 2015); U.S. Pat. No. 7,394,392Nonprecedential · per curiam · from N.D. Ill. (Hart, S.J.) | Using one or more expert systems to screen vehicle and equipment operators for impairment, test them selectively, and control the equipment. | Ineligible | The claims are directed to testing operators for impairment, and neither the claims nor the specification explain how the expert system works, leaving only existing vehicle equipment and commercially available processors. |
What is included
Included: written opinions of the United States Court of Appeals for the Federal Circuit, precedential and nonprecedential, issued since Alice Corp. v. CLS Bank International (June 19, 2014), that decided eligibility under 35 U.S.C. § 101 for claims built around machine learning, a trained or learned model, a statistical model used for prediction, or a rule-based expert system. Appeals from district courts and from the Patent Trial and Appeal Board are both included.
Two inclusions call for a word of explanation. Trustees of Columbia University v. Gen Digital is included because its claims recite an anomaly-detection model, although the court did not describe the case as one about machine learning. The two In re Board of Trustees of the Leland Stanford Junior University decisions are included because their claims recite hidden Markov models, which are statistical models used for prediction.
Not included: affirmances without an opinion under Federal Circuit Rule 36, such as PurePredictive, Inc. v. H2O.AI, Inc., No. 2017-2544 (Fed. Cir. Nov. 7, 2018), which affirmed a district court ruling that machine learning ensemble claims were ineligible; decisions that mention artificial intelligence only in a party's name; and decisions on other grounds, such as Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022), on whether an AI system can be named as an inventor.
This table lists every such opinion identified as of the date above. If you know of one that is missing, please write to jon@burbagelaw.com.
Suggested citation: Jon Burbage, Federal Circuit Decisions on Artificial Intelligence Patent Eligibility, Burbage Law (updated Oct. 26, 2026), https://burbagelaw.com/ai-eligibility.html
Summaries are general information, not legal advice, and do not create an attorney-client relationship. Read the opinion before relying on any decision.